Abstract Self-similarity refers to the image prior widely used in image restoration algorithms that small but similar patterns tend to occur at different locations and scales. However, recent advanced deep convolutional neural network-based methods for image restoration do not take full advantage of self-similarities by relying on self-attention neural modules that only process information at the same scale. To solve this problem, we present a novel Pyramid Attention module for image restoration, which captures long-range feature correspondences from a multi-scale feature pyramid. Inspired by the fact that corruptions, such as noise or compression artifacts, drop drastically at coarser image scales, our attention module is designed to be able to borrow clean signals from their “clean” correspondences at the coarser levels. The proposed pyramid attention module is a generic building block that can be flexibly integrated into various neural architectures. Its effectiveness is validated through extensive experiments on multiple image restoration tasks: image denoising, demosaicing, compression artifact reduction, and super resolution. Without any bells and whistles, our PANet (pyramid attention module with simple network backbones) can produce state-of-the-art results with superior accuracy and visual quality. Our code is available at https://github.com/SHI-Labs/Pyramid-Attention-Networks

Contextual information is vital in visual understanding problems, such as semantic segmentation and object detection. We propose a criss-cross network (CCNet) for obtaining full-image contextual information in a very effective and efficient way. Concretely, for each pixel, a novel criss-cross attention module harvests the contextual information of all the pixels on its criss-cross path. By taking a further recurrent operation, each pixel can finally capture the full-image dependencies. Besides, a category consistent loss is proposed to enforce the criss-cross attention module to produce more discriminative features. Overall, CCNet is with the following merits: 1) GPU memory friendly. Compared with the non-local block, the proposed recurrent criss-cross attention module requires 11× less GPU memory usage. 2) High computational efficiency. The recurrent criss-cross attention significantly reduces FLOPs by about 85 percent of the non-local block. 3) The state-of-the-art performance. We conduct extensive experiments on semantic segmentation benchmarks including Cityscapes, ADE20K, human parsing benchmark LIP, instance segmentation benchmark COCO, video segmentation benchmark CamVid. In particular, our CCNet achieves the mIoU scores of 81.9, 45.76 and 55.47 percent on the Cityscapes test set, the ADE20K validation set and the LIP validation set respectively, which are the new state-of-the-art results. The source codes are available at https://github.com/speedinghzl/CCNethttps://github.com/speedinghzl/CCNet.

Aggregating features in terms of different convolutional blocks or contextual embeddings has been proven to be an effective way to strengthen feature representations for semantic segmentation. However, most of the current popular network architectures tend to ignore the misalignment issues during the feature aggregation process caused by step-by-step downsampling operations and indiscriminate contextual information fusion. In this paper, we explore the principles in addressing such feature misalignment issues and inventively propose Feature-Aligned Segmentation Networks (AlignSeg). AlignSeg consists of two primary modules, i.e., the Aligned Feature Aggregation (AlignFA) module and the Aligned Context Modeling (AlignCM) module. First, AlignFA adopts a simple learnable interpolation strategy to learn transformation offsets of pixels, which can effectively relieve the feature misalignment issue caused by multi-resolution feature aggregation. Second, with the contextual embeddings in hand, AlignCM enables each pixel to choose private custom contextual information adaptively, making the contextual embeddings be better aligned. We validate the effectiveness of our AlignSeg network with extensive experiments on Cityscapes and ADE20K, achieving new state-of-the-art mIoU scores of 82.6 and 45.95 percent, respectively. Our source code is available at https://github.com/speedinghzl/AlignSeg.

Video instance segmentation is a complex task in which we need to detect, segment, and track each object for any given video. Previous approaches only utilize single-frame features for the detection, segmentation, and tracking of objects and they suffer in the video scenario due to several distinct challenges such as motion blur and drastic appearance change. To eliminate ambiguities introduced by only using single-frame features, we propose a novel comprehensive feature aggregation approach (CompFeat) to refine features atboth frame-level and object-level with temporal and spatial context information. The aggregation process is carefully designed with a new attention mechanism which significantly increases the discriminative power of the learned features. We further improve the tracking capability of our model through a siamese design by incorporating both feature similarities and spatial similarities. Experiments conducted on the YouTube-VIS dataset validate the effectiveness of proposed CompFeat.

Existing deep learning-based models for image captioning typically consist of an image encoder to extract visual features and a language model decoder, an architecture that has shown promising results in single high-level sentence generation. However, only the word-level guiding signal is available when the image encoder is optimized to extract visual features. The inconsistency between the parallel extraction of visual features and sequential text supervision limits its success when the length of the generated text is long (more than 50 words). We propose a new module, called the Text Embedding Bank (TEB), to address this problem for image paragraph captioning. This module uses the paragraph vector model to learn fixed-length feature representations from a variable-length paragraph. We refer to the fixed-length feature as the TEB. This TEB module plays two roles to benefit paragraph captioning performance. First, it acts as a form of global and coherent deep supervision to regularize visual feature extraction in the image encoder. Second, it acts as a distributed memory to provide features of the whole paragraph to the language model, which alleviates the long-term dependency problem. Adding this module to two existing state-of-the-art methods achieves a new state-of-the-art result on the paragraph captioning Stanford Visual Genome dataset.

We present any-precision deep neural networks (DNNs), which are trained with a new method that allows the learned DNNs to be flexible in numerical precision during inference. The same model in runtime can be flexibly and directly set to different bit-widths, by truncating the least significant bits, to support dynamic speed and accuracy trade-off. When all layers are set to low-bits, we show that the model achieved accuracy comparable to dedicated models trained at the same precision. This nice property facilitates flexible deployment of deep learning models in real-world applications, where in practice trade-offs between model accuracy and runtime efficiency are often sought. Previous literature presents solutions to train models at each individual fixed efficiency/accuracy trade-off point. But how to produce a model flexible in runtime precision is largely unexplored. When the demand of efficiency/accuracy trade-off varies from time to time or even dynamically changes in runtime, it is infeasible to re-train models accordingly, and the storage budget may forbid keeping multiple models. Our proposed framework achieves this flexibility without performance degradation. More importantly, we demonstrate that this achievement is agnostic to model architectures and applicable to multiple vision tasks. Our code is released at https://github.com/SHI-Labs/Any-Precision-DNNs.

While scale-invariant modeling has substantially boosted the performance of visual recognition tasks, it remains largely under-explored in deep networks based image restoration. Naively applying those scale-invariant techniques (e.g., multi-scale testing, random-scale data augmentation) to image restoration tasks usually leads to inferior performance. In this paper, we show that properly modeling scale-invariance into neural networks can bring significant benefits to image restoration performance. Inspired from spatial-wise convolution for shift-invariance, “scale-wise convolution” is proposed to convolve across multiple scales for scale-invariance. In our scale-wise convolutional network (SCN), we first map the input image to the feature space and then build a feature pyramid representation via bi-linear down-scaling progressively. The feature pyramid is then passed to a residual network with scale-wise convolutions. The proposed scale-wise convolution learns to dynamically activate and aggregate features from different input scales in each residual building block, in order to exploit contextual information on multiple scales. In experiments, we compare the restoration accuracy and parameter efficiency among our model and many different variants of multi-scale neural networks. The proposed network with scale-wise convolution achieves superior performance in multiple image restoration tasks including image super-resolution, image denoising and image compression artifacts removal. Code and models are available at: https://github.com/ychfan/scn_sr.

Talking face synthesis has been widely studied in either appearance-based or warping-based methods. Previous works mostly utilize single face image as a source, and generate novel facial animations by merging other person's facial features. However, some facial regions like eyes or teeth, which may be hidden in the source image, can not be synthesized faithfully and stably. In this paper, We present a landmark driven two-stream network to generate faithful talking facial animation, in which more facial details are created, preserved and transferred from multiple source images instead of a single one. Specifically, we propose a network consisting of a learning and fetching stream. The fetching sub-net directly learns to attentively warp and merge facial regions from five source images of distinctive landmarks, while the learning pipeline renders facial organs from the training face space to compensate. Compared to baseline algorithms, extensive experiments demonstrate that the proposed method achieves a higher performance both quantitatively and qualitatively. Codes are at https://github.com/kgu3/FLNet_AAAI2020.

Discriminative learning based image denoisers have achieved promising performance on synthetic noises such as Additive White Gaussian Noise (AWGN). The synthetic noises adopted in most previous work are pixel-independent, but real noises are mostly spatially/channel-correlated and spatially/channel-variant. This domain gap yields unsatisfied performance on images with real noises if the model is only trained with AWGN. In this paper, we propose a novel approach to boost the performance of a real image denoiser which is trained only with synthetic pixel-independent noise data dominated by AWGN. First, we train a deep model that consists of a noise estimator and a denoiser with mixed AWGN and Random Value Impulse Noise (RVIN). We then investigate Pixel-shuffle Down-sampling (PD) strategy to adapt the trained model to real noises. Extensive experiments demonstrate the effectiveness and generalization of the proposed approach. Notably, our method achieves state-of-the-art performance on real sRGB images in the DND benchmark among models trained with synthetic noises. Codes are available at https://github.com/yzhouas/PD-Denoising-pytorch.

In this work, we introduce Panoptic-DeepLab, a simple, strong, and fast system for panoptic segmentation, aiming to establish a solid baseline for bottom-up methods that can achieve comparable performance of two-stage methods while yielding fast inference speed. In particular, Panoptic-DeepLab adopts the dual-ASPP and dual-decoder structures specific to semantic, and instance segmentation, respectively. The semantic segmentation branch is the same as the typical design of any semantic segmentation model (e.g., DeepLab), while the instance segmentation branch is class-agnostic, involving a simple instance center regression. As a result, our single Panoptic-DeepLab simultaneously ranks first at all three Cityscapes benchmarks, setting the new state-of-art of 84.2% mIoU, 39.0% AP, and 65.5% PQ on test set. Additionally, equipped with MobileNetV3, Panoptic-DeepLab runs nearly in real-time with a single 1025x2049 image (15.8 frames per second), while achieving a competitive performance on Cityscapes (54.1 PQ% on test set). On Mapillary Vistas test set, our ensemble of six models attains 42.7% PQ, outperforming the challenge winner in 2018 by a healthy margin of 1.5%. Finally, our Panoptic-DeepLab also performs on par with several top-down approaches on the challenging COCO dataset. For the first time, we demonstrate a bottom-up approach could deliver state-of-the-art results on panoptic segmentation.

Bottom-up human pose estimation methods have difficulties in predicting the correct pose for small persons due to challenges in scale variation. In this paper, we present HigherHRNet: a novel bottom-up human pose estimation method for learning scale-aware representations using high-resolution feature pyramids. Equipped with multi-resolution supervision for training and multi-resolution aggregation for inference, the proposed approach is able to solve the scale variation challenge in bottom-up multi-person pose estimation and localize keypoints more precisely, especially for small person. The feature pyramid in HigherHRNet consists of feature map outputs from HRNet and upsampled higher-resolution outputs through a transposed convolution. HigherHRNet outperforms the previous best bottom-up method by 2.5% AP for medium person on COCO test-dev, showing its effectiveness in handling scale variation. Furthermore, HigherHRNet achieves new state-of-the-art result on COCO test-dev (70.5% AP) without using refinement or other post-processing techniques, surpassing all existing bottom-up methods. HigherHRNet even surpasses all top-down methods on CrowdPose test (67.6% AP), suggesting its robustness in crowded scene.

The success of deep learning in visual recognition tasks has driven advancements in multiple fields of research. Particularly, increasing attention has been drawn towards its application in agriculture. Nevertheless, while visual pattern recognition on farmlands carries enormous economic values, little progress has been made to merge computer vision and crop sciences due to the lack of suitable agricultural image datasets. Meanwhile, problems in agriculture also pose new challenges in computer vision. For example, semantic segmentation of aerial farmland images requires inference over extremely large-size images with extreme annotation sparsity. These challenges are not present in most of the common object datasets, and we show that they are more challenging than many other aerial image datasets. To encourage research in computer vision for agriculture, we present Agriculture-Vision: a large-scale aerial farmland image dataset for semantic segmentation of agricultural patterns. We collected 94,986 high-quality aerial images from 3,432 farmlands across the US, where each image consists of RGB and Near-infrared (NIR) channels with resolution as high as 10 cm per pixel. We annotate nine types of field anomaly patterns that are most important to farmers. As a pilot study of aerial agricultural semantic segmentation, we perform comprehensive experiments using popular semantic segmentation models; we also propose an effective model designed for aerial agricultural pattern recognition. Our experiments demonstrate several challenges Agriculture-Vision poses to both the computer vision and agriculture communities. Future versions of this dataset will include even more aerial images, anomaly patterns and image channels.

Deep convolution-based single image super-resolution (SISR) networks embrace the benefits of learning from large-scale external image resources for local recovery, yet most existing works have ignored the long-range feature-wise similarities in natural images. Some recent works have successfully leveraged this intrinsic feature correlation by exploring non-local attention modules. However, none of the current deep models have studied another inherent property of images: cross-scale feature correlation. In this paper, we propose the first Cross-Scale Non-Local (CS-NL) attention module with integration into a recurrent neural network. By combining the new CS-NL prior with local and in-scale non-local priors in a powerful recurrent fusion cell, we can find more cross-scale feature correlations within a single low-resolution (LR) image. The performance of SISR is significantly improved by exhaustively integrating all possible priors. Extensive experiments demonstrate the effectiveness of the proposed CS-NL module by setting new state-of-the-arts on multiple SISR benchmarks.

We consider the problem of unsupervised domain adaptation for semantic segmentation by easing the domain shift between the source domain (synthetic data) and the target domain (real data) in this work. State-of-the-art approaches prove that performing semantic-level alignment is helpful in tackling the domain shift issue. Based on the observation that stuff categories usually share similar appearances across images of different domains while things (i.e. object instances) have much larger differences, we propose to improve the semantic-level alignment with different strategies for stuff regions and for things: 1) for the stuff categories, we generate feature representation for each class and conduct the alignment operation from the target domain to the source domain; 2) for the thing categories, we generate feature representation for each individual instance and encourage the instance in the target domain to align with the most similar one in the source domain. In this way, the individual differences within thing categories will also be considered to alleviate over-alignment. In addition to our proposed method, we further reveal the reason why the current adversarial loss is often unstable in minimizing the distribution discrepancy and show that our method can help ease this issue by minimizing the most similar stuff and instance features between the source and the target domains. We conduct extensive experiments in two unsupervised domain adaptation tasks, i.e. GTA5 → Cityscapes and SYNTHIA → Cityscapes, and achieve the new state-of-the-art segmentation accuracy.

Motion estimation of cardiac MRI videos is crucial for the evaluation of human heart anatomy and function. Recent researches show promising results with deep learning-based methods. In clinical deployment, however, they suffer dramatic performance drops due to mismatched distributions between training and testing datasets, commonly encountered in the clinical environment. On the other hand, it is arguably impossible to collect all representative datasets and to train a universal tracker before deployment. In this context, we proposed a novel fast online adaptive learning (FOAL) framework: an online gradient descent based optimizer that is optimized by a meta-learner. The meta-learner enables the online optimizer to perform a fast and robust adaptation. We evaluated our method through extensive experiments on two public clinical datasets. The results showed the superior performance of FOAL in accuracy compared to the offline-trained tracking method. On average, the FOAL took only 0.4 second per video for online optimization.

Inspired by the robustness and efficiency of sparse representation in sparse coding based image restoration models, we investigate the sparsity of neurons in deep networks. Our method structurally enforces sparsity constraints upon hidden neurons. The sparsity constraints are favorable for gradient-based learning algorithms and attachable to convolution layers in various networks. Sparsity in neurons enables computation saving by only operating on non-zero components without hurting accuracy. Meanwhile, our method can magnify representation dimensionality and model capacity with negligible additional computation cost. Experiments show that sparse representation is crucial in deep neural networks for multiple image restoration tasks, including image super-resolution, image denoising, and image compression artifacts removal.

Recently image-to-image translation has attracted significant interests in the literature, starting from the successful use of the generative adversarial network (GAN), to the introduction of cyclic constraint, to extensions to multiple domains. However, in existing approaches, there is no guarantee that the mapping between two image domains is unique or one-to-one. Here we propose a self-inverse network learning approach for unpaired image-to-image translation. Building on top of CycleGAN, we learn a self-inverse function by simply augmenting the training samples by swapping inputs and outputs during training and with separated cycle consistency loss for each mapping direction. The outcome of such learning is a proven one-to-one mapping function. Our extensive experiments on a variety of datasets, including cross-modal medical image synthesis, object transfiguration, and semantic labeling, consistently demonstrate clear improvement over the CycleGAN method both qualitatively and quantitatively. Especially our proposed method reaches the state-of-the-art result on the cityscapes benchmark dataset for the label to photo un-paired directional image translation.

One-shot image semantic segmentation poses a challenging task of recognizing the object regions from unseen categories with only one annotated example as supervision. In this article, we propose a simple yet effective similarity guidance network to tackle the one-shot (SG-One) segmentation problem. We aim at predicting the segmentation mask of a query image with the reference to one densely labeled support image of the same category. To obtain the robust representative feature of the support image, we first adopt a masked average pooling strategy for producing the guidance features by only taking the pixels belonging to the support image into account. We then leverage the cosine similarity to build the relationship between the guidance features and features of pixels from the query image. In this way, the possibilities embedded in the produced similarity maps can be adopted to guide the process of segmenting objects. Furthermore, our SG-One is a unified framework that can efficiently process both support and query images within one network and be learned in an end-to-end manner. We conduct extensive experiments on Pascal VOC 2012. In particular, our SG-One achieves the mIoU score of 46.3%, surpassing the baseline methods.

In this paper, we propose a deep CNN to tackle the image restoration problem by learning formatted information. Previous deep learning based methods directly learn the mapping from corrupted images to clean images, and may suffer from the gradient exploding/vanishing problems of deep neural networks. We propose to address the image restoration problem by learning the structured details and recovering the latent clean image together, from the shared information between the corrupted image and the latent image. In addition, instead of learning the pure difference (corruption), we propose to add a residual formatting layer and an adversarial block to format the information to structured one, which allows the network to converge faster and boosts the performance. Furthermore, we propose a cross-level loss net to ensure both pixel-level accuracy and semantic-level visual quality. Evaluations on public datasets show that the proposed method performs favorably against existing approaches quantitatively and qualitatively.

Image denoising and high-level vision tasks are usually handled independently in the conventional practice of computer vision, and their connection is fragile. In this paper, we cope with the two jointly and explore the mutual influence between them with the focus on two questions, namely (1) how image denoising can help improving high-level vision tasks, and (2) how the semantic information from high-level vision tasks can be used to guide image denoising. First for image denoising we propose a convolutional neural network in which convolutions are conducted in various spatial resolutions via downsampling and upsampling operations in order to fuse and exploit contextual information on different scales. Second we propose a deep neural network solution that cascades two modules for image denoising and various high-level tasks, respectively, and use the joint loss for updating only the denoising network via backpropagation. We experimentally show that on one hand, the proposed denoiser has the generality to overcome the performance degradation of different high-level vision tasks. On the other hand, with the guidance of high-level vision information, the denoising network produces more visually appealing results. Extensive experiments demonstrate the benefit of exploiting image semantics simultaneously for image denoising and highlevel vision tasks via deep learning. The code is available online: https://github.com/Ding-Liu/DeepDenoising.

In this work, we propose a novel Spatial-Temporal Attention (STA) approach to tackle the large-scale person reidentification task in videos. Different from the most existing methods, which simply compute representations of video clips using frame-level aggregation (e.g. average pooling), the proposed STA adopts a more effective way for producing robust clip-level feature representation. Concretely, our STA fully exploits those discriminative parts of one target person in both spatial and temporal dimensions, which results in a 2-D attention score matrix via inter-frame regularization to measure the importances of spatial parts across different frames. Thus, a more robust clip-level feature representation can be generated according to a weighted sum operation guided by the mined 2-D attention score matrix. In this way, the challenging cases for video-based person re-identification such as pose variation and partial occlusion can be well tackled by the STA. We conduct extensive experiments on two large-scale benchmarks, i.e. MARS and DukeMTMCVideoReID. In particular, the mAP reaches 87.7% on MARS, which significantly outperforms the state-of-the-arts with a large margin of more than 11.6%.

Despite the remarkable progress in person re-identification (Re-ID), such approaches still suffer from the failure cases where the discriminative body parts are missing. To mitigate this type of failure, we propose a simple yet effective Horizontal Pyramid Matching (HPM) approach to fully exploit various partial information of a given person, so that correct person candidates can be identified even if some key parts are missing. With HPM, we make the following contributions to produce more robust feature representations for the Re-ID task: 1) we learn to classify using partial feature representations at different horizontal pyramid scales, which successfully enhance the discriminative capabilities of various person parts; 2) we exploit average and max pooling strategies to account for person-specific discriminative information in a global-local manner. To validate the effectiveness of our proposed HPM method, extensive experiments are conducted on three popular datasets including Market-1501, DukeMTMCReID and CUHK03. Respectively, we achieve mAP scores of 83.1%, 74.5% and 59.7% on these challenging benchmarks, which are the new state-of-the-arts.

Semantic scene parsing is suffering from the fact that pixellevel annotations are hard to be collected. To tackle this issue, we propose a Point-based Distance Metric Learning (PDML) in this paper. PDML does not require dense annotated masks and only leverages several labeled points that are much easier to obtain to guide the training process. Concretely, we leverage semantic relationship among the annotated points by encouraging the feature representations of the intra- and intercategory points to keep consistent, i.e. points within the same category should have more similar feature representations compared to those from different categories. We formulate such a characteristic into a simple distance metric loss, which collaborates with the point-wise cross-entropy loss to optimize the deep neural networks. Furthermore, to fully exploit the limited annotations, distance metric learning is conducted across different training images instead of simply adopting an image-dependent manner. We conduct extensive experiments on two challenging scene parsing benchmarks of PASCALContext and ADE 20K to validate the effectiveness of our PDML, and competitive mIoU scores are achieved.

Discriminative learning based denoising model trained with Additive White Gaussian Noise (AWGN) performs well on synthesized noise. However, realistic noise can be spatialvariant, signal-dependent and a mixture of complicated noises. In this paper, we explore multiple strategies for applying an AWGN-based denoiser to realistic noise. Specifically, we trained a deep network integrating noise estimating and denoiser with mixed Gaussian (AWGN) and Random Value Impulse Noise (RVIN). To adapt the model to realistic noises, we investigated multi-channel, multi-scale and super-resolution approaches. Our preliminary results demonstrated the effectiveness of the newly-proposed noise model and adaptation strategies.

It has been shown that jointly reasoning the 2D appearance and 3D information from RGB-D domains is beneficial to indoor scene semantic segmentation. However, most existing approaches require accurate depth map as input to segment the scene which severely limits their applications. In this paper, we propose to jointly infer the semantic and depth information by distilling geometry-aware embedding to eliminate such strong constraint while still exploiting the helpful depth domain information. In addition, we use this learned embedding to improve the quality of semantic segmentation, through a proposed geometry-aware propagation framework followed by several multi-level skip feature fusion blocks. By decoupling the single task prediction network into two joint tasks of semantic segmentation and geometry embedding learning, together with the proposed information propagation and feature fusion architecture, our method is shown to perform favorably against state-of-the-art methods for semantic segmentation on publicly available challenging indoor datasets.

Unpaired Image-to-image Translation is a new rising and challenging vision problem that aims to learn a mapping between unaligned image pairs in diverse domains. Recent advances in this field like MUNIT and DRIT mainly focus on disentangling content and style/attribute from a given image first, then directly adopting the global style to guide the model to synthesize new domain images. However, this kind of approaches severely incurs contradiction if the target domain images are content-rich with multiple discrepant objects. In this paper, we present a simple yet effective instance-aware image-to-image translation approach (INIT), which employs the fine-grained local (instance) and global styles to the target image spatially. The proposed INIT exhibits three import advantages: (1) the instance-level objective loss can help learn a more accurate reconstruction and incorporate diverse attributes of objects; (2) the styles used for target domain of local/global areas are from corresponding spatial regions in source domain, which intuitively is a more reasonable mapping; (3) the joint training process can benefit both fine and coarse granularity and incorporates instance information to improve the quality of global translation. We also collect a large-scale benchmark for the new instance-level translation task. We observe that our synthetic images can even benefit real-world vision tasks like generic object detection.

Multi-scale context module and single-stage encoder-decoder structure are commonly employed for semantic segmentation. The multi-scale context module refers to the operations to aggregate feature responses from a large spatial extent, while the single-stage encoder-decoder structure encodes the high-level semantic information in the encoder path and recovers the boundary information in the decoder path. In contrast, multi-stage encoder-decoder networks have been widely used in human pose estimation and show superior performance than their single-stage counterpart. However, few efforts have been attempted to bring this effective design to semantic segmentation. In this work, we propose a Semantic Prediction Guidance (SPG) module which learns to re-weight the local features through the guidance from pixel-wise semantic prediction. We find that by carefully re-weighting features across stages, a two-stage encoder-decoder network coupled with our proposed SPG module can significantly outperform its one-stage counterpart with similar parameters and computations. Finally, we report experimental results on the semantic segmentation benchmark Cityscapes, in which our SPGNet attains 81.1% on the test set using only 'fine' annotations.

Domain adaptation in person re-identification (re-ID) has always been a challenging task. In this work, we explore how to harness the similar natural characteristics existing in the samples from the target domain for learning to conduct person re-ID in an unsupervised manner. Concretely, we propose a Self-similarity Grouping (SSG) approach, which exploits the potential similarity (from the global body to local parts) of unlabeled samples to build multiple clusters from different views automatically. These independent clusters are then assigned with labels, which serve as the pseudo identities to supervise the training process. We repeatedly and alternatively conduct such a grouping and training process until the model is stable. Despite the apparent simplify, our SSG outperforms the state-of-the-arts by more than 4.6% (DukeMTMC→Market1501) and 4.4% (Market1501→DukeMTMC) in mAP, respectively. Upon our SSG, we further introduce a clustering-guided semisupervised approach named SSG ++ to conduct the one-shot domain adaption in an open set setting (i.e. the number of independent identities from the target domain is unknown). Without spending much effort on labeling, our SSG ++ can further promote the mAP upon SSG by 10.7% and 6.9%, respectively. Our Code is available at: https://github.com/OasisYang/SSG .

Slimmable networks are a family of neural networks that can instantly adjust the runtime width. The width can be chosen from a predefined widths set to adaptively optimize accuracy-efficiency trade-offs at runtime. In this work, we propose a systematic approach to train universally slimmable networks (US-Nets), extending slimmable networks to execute at arbitrary width, and generalizing to networks both with and without batch normalization layers. We further propose two improved training techniques for US-Nets, named the sandwich rule and inplace distillation, to enhance training process and boost testing accuracy. We show improved performance of universally slimmable MobileNet v1 and MobileNet v2 on ImageNet classification task, compared with individually trained ones and 4-switch slimmable network baselines. We also evaluate the proposed US-Nets and improved training techniques on tasks of image super-resolution and deep reinforcement learning. Extensive ablation experiments on these representative tasks demonstrate the effectiveness of our proposed methods. Our discovery opens up the possibility to directly evaluate FLOPs-Accuracy spectrum of network architectures. Code and models are available at: \url{https://github.com/JiahuiYu/slimmable_networks}.

We present a generative image inpainting system to complete images with free-form mask and guidance. The system is based on gated convolutions learned from millions of images without additional labelling efforts. The proposed gated convolution solves the issue of vanilla convolution that treats all input pixels as valid ones, generalizes partial convolution by providing a learnable dynamic feature selection mechanism for each channel at each spatial location across all layers. Moreover, as free-form masks may appear anywhere in images with any shape, global and local GANs designed for a single rectangular mask are not applicable. Thus, we also present a patch-based GAN loss, named SN-PatchGAN, by applying spectral-normalized discriminator on dense image patches. SN-PatchGAN is simple in formulation, fast and stable in training. Results on automatic image inpainting and user-guided extension demonstrate that our system generates higher-quality and more flexible results than previous methods. Our system helps user quickly remove distracting objects, modify image layouts, clear watermarks and edit faces. Code, demo and models are available at: \url{https://github.com/JiahuiYu/generative_inpainting}.

A good representation for arbitrarily complicated data should have the capability of semantic generation, clustering and reconstruction. Previous research has already achieved impressive performance on either one. This paper aims at learning a disentangled representation effective for all of them in an unsupervised way. To achieve all the three tasks together, we learn the forward and inverse mapping between data and representation on the basis of a symmetric adversarial process. In theory, we minimize the upper bound of the two conditional entropy loss between the latent variables and the observations together to achieve the cycle consistency. The newly proposed RepGAN is tested on MNIST, fashionMNIST, CelebA, and SVHN datasets to perform unsupervised classification, generation and reconstruction tasks. The result demonstrates that RepGAN is able to learn a useful and competitive representation. To the author's knowledge, our work is the first one to achieve both a high unsupervised classification accuracy and low reconstruction error on MNIST.

3D human pose estimation from monocular images has become a heated area in computer vision recently. For years, most deep neural network based practices have adopted either an end-to-end approach, or a two-stage approach. An end-to-end network typically estimates 3D human poses directly from 2D input images, but it suffers from the shortage of 3D human pose data. It is also obscure to know if the inaccuracy stems from limited visual under-standing or 2D-to-3D mapping. Whereas a two-stage directly lifts those 2D keypoint outputs to the 3D space, after utilizing an existing network for 2D keypoint detections. However, they tend to ignore some useful contextual hints from the 2D raw image pixels. In this paper, we introduce a two-stage architecture that can eliminate the main disadvantages of both these approaches. During the first stage we use an existing state-of-the-art detector to estimate 2D poses. To add more con-textual information to help lifting 2D poses to 3D poses, we propose 3D Part Affinity Fields (3D-PAFs). We use 3D-PAFs to infer 3D limb vectors, and combine them with 2D poses to regress the 3D coordinates. We trained and tested our proposed framework on Human3.6M, the most popular 3D human pose benchmark dataset. Our approach achieves the state-of-the-art performance, which proves that with right selections of contextual information, a simple regression model can be very powerful in estimating 3D poses.

Visual recognition under adverse conditions is a very important and challenging problem of high practical value, due to the ubiquitous existence of quality distortions during image acquisition, transmission, or storage. While deep neural networks have been extensively exploited in the techniques of low-quality image restoration and high-quality image recognition tasks respectively, few studies have been done on the important problem of recognition from very low-quality images. This paper proposes a deep learning based framework for improving the performance of image and video recognition models under adverse conditions, using robust adverse pre-training or its aggressive variant. The robust adverse pre-training algorithms leverage the power of pre-training and generalizes conventional unsupervised pre-training and data augmentation methods. We further develop a transfer learning approach to cope with real-world datasets of unknown adverse conditions. The proposed framework is comprehensively evaluated on a number of image and video recognition benchmarks, and obtains significant performance improvements under various single or mixed adverse conditions. Our visualization and analysis further add to the explainability of results.

Visual recognition from very low-quality images is an extremely challenging task with great practical values. While deep networks have been extensively applied to low-quality image restoration and high-quality image recognition tasks respectively, few works have been done on the important problem of recognition from very low-quality images.This paper presents a degradation-robust pre-training approach on improving deep learning models towards this direction. Extensive experiments on different datasets validate the effectiveness of our proposed method.

Advances in image super-resolution (SR) have recently benefited significantly from rapid developments in deep neural networks. Inspired by these recent discoveries, we note that many state-of-the-art deep SR architectures can be reformulated as a single-state recurrent neural network (RNN) with finite unfoldings. In this paper, we explore new structures for SR based on this compact RNN view, leading us to a dual-state design, the Dual-State Recurrent Network (DSRN). Compared to its single-state counterparts that operate at a fixed spatial resolution, DSRN exploits both low-resolution (LR) and high-resolution (HR) signals jointly. Recurrent signals are exchanged between these states in both directions (both LR to HR and HR to LR) via delayed feedback. Extensive quantitative and qualitative evaluations on benchmark datasets and on a recent challenge demonstrate that the proposed DSRN performs favorably against state-of-the-art algorithms in terms of both memory consumption and predictive accuracy. The code for our method is publicly available1.

Despite the remarkable progress, weakly supervised segmentation approaches are still inferior to their fully supervised counterparts. We obverse the performance gap mainly comes from their limitation on learning to produce high-quality dense object localization maps from image-level supervision. To mitigate such a gap, we revisit the dilated convolution [1] and reveal how it can be utilized in a novel way to effectively overcome this critical limitation of weakly supervised segmentation approaches. Specifically, we find that varying dilation rates can effectively enlarge the receptive fields of convolutional kernels and more importantly transfer the surrounding discriminative information to non-discriminative object regions, promoting the emergence of these regions in the object localization maps. Then, we design a generic classification network equipped with convolutional blocks of different dilated rates. It can produce dense and reliable object localization maps and effectively benefit both weakly- and semi- supervised semantic segmentation. Despite the apparent simplicity, our proposed approach obtains superior performance over state-of-the-arts. In particular, it achieves 60.8% and 67.6% mIoU scores on Pascal VOC 2012 test set in weakly- (only image-level labels are available) and semi- (1,464 segmentation masks are available) supervised settings, which are the new state-of-the-arts.

Recent deep learning based approaches have shown promising results for the challenging task of inpainting large missing regions in an image. These methods can generate visually plausible image structures and textures, but often create distorted structures or blurry textures inconsistent with surrounding areas. This is mainly due to ineffectiveness of convolutional neural networks in explicitly borrowing or copying information from distant spatial locations. On the other hand, traditional texture and patch synthesis approaches are particularly suitable when it needs to borrow textures from the surrounding regions. Motivated by these observations, we propose a new deep generative model-based approach which can not only synthesize novel image structures but also explicitly utilize surrounding image features as references during network training to make better predictions. The model is a feedforward, fully convolutional neural network which can process images with multiple holes at arbitrary locations and with variable sizes during the test time. Experiments on multiple datasets including faces (CelebA, CelebA-HQ), textures (DTD) and natural images (ImageNet, Places2) demonstrate that our proposed approach generates higher-quality inpainting results than existing ones. Code, demo and models are available at: https://github.com/JiahuiYu/generative_inpainting.

In this work, we propose Adversarial Complementary Learning (ACoL) to automatically localize integral objects of semantic interest with weak supervision. We first mathematically prove that class localization maps can be obtained by directly selecting the class-specific feature maps of the last convolutional layer, which paves a simple way to identify object regions. We then present a simple network architecture including two parallel-classifiers for object localization. Specifically, we leverage one classification branch to dynamically localize some discriminative object regions during the forward pass. Although it is usually responsive to sparse parts of the target objects, this classifier can drive the counterpart classifier to discover new and complementary object regions by erasing its discovered regions from the feature maps. With such an adversarial learning, the two parallel-classifiers are forced to leverage complementary object regions for classification and can finally generate integral object localization together. The merits of ACoL are mainly two-fold: 1) it can be trained in an end-to-end manner; 2) dynamically erasing enables the counterpart classifier to discover complementary object regions more effectively. We demonstrate the superiority of our ACoL approach in a variety of experiments. In particular, the Top-1 localization error rate on the ILSVRC dataset is 45.14%, which is the new state-of-the-art.

Face detection is a well-explored problem. Many challenges on face detectors like extreme pose, illumination, low resolution and small scales are studied in the previous work. However, previous proposed models are mostly trained and tested on good-quality images which are not always the case for practical applications like surveillance systems. In this paper, we first review the current state-of-the-art face detectors and their performance on benchmark dataset FDDB, and compare the design protocols of the algorithms. Secondly, we investigate their performance degradation while testing on low-quality images with different levels of blur, noise, and contrast. Our results demonstrate that both hand-crafted and deep-learning based face detectors are not robust enough for low-quality images. It inspires researchers to produce more robust design for face detection in the wild.

Conventionally, image denoising and high-level vision tasks are handled separately in computer vision. In this paper, we cope with the two jointly and explore the mutual influence between them. First we propose a convolutional neural network for image denoising which achieves the state-of-the-art performance. Second we propose a deep neural network solution that cascades two modules for image denoising and various high-level tasks, respectively, and use the joint loss for updating only the denoising network via back-propagation. We demonstrate that on one hand, the proposed denoiser has the generality to overcome the performance degradation of different high-level vision tasks. On the other hand, with the guidance of high-level vision information, the denoising network can generate more visually appealing results. To the best of our knowledge, this is the first work investigating the benefit of exploiting image semantics simultaneously for image denoising and high-level vision tasks via deep learning.

Many classic methods have shown non-local self-similarity in natural images to be an effective prior for image restoration. However, it remains unclear and challenging to make use of this intrinsic property via deep networks. In this paper, we propose a non-local recurrent network (NLRN) as the first attempt to incorporate non-local operations into a recurrent neural network (RNN) for image restoration. The main contributions of this work are: (1) Unlike existing methods that measure self-similarity in an isolated manner, the proposed non-local module can be flexibly integrated into existing deep networks for end-to-end training to capture deep feature correlation between each location and its neighborhood. (2) We fully employ the RNN structure for its parameter efficiency and allow deep feature correlation to be propagated along adjacent recurrent states. This new design boosts robustness against inaccurate correlation estimation due to severely degraded images. (3) We show that it is essential to maintain a confined neighborhood for computing deep feature correlation given degraded images. This is in contrast to existing practice that deploys the whole image. Extensive experiments on both image denoising and super-resolution tasks are conducted. Thanks to the recurrent non-local operations and correlation propagation, the proposed NLRN achieves superior results to state-of-the-art methods with many fewer parameters.

In this paper, we propose to exploit the interactions between non-associable tracklets to facilitate multi-object tracking. We introduce two types of tracklet interactions, close interaction and distant interaction. The close interaction imposes physical constraints between two temporally overlapping tracklets and more importantly, allows us to learn local classifiers to distinguish targets that are close to each other in the spatiotemporal domain. The distant interaction, on the other hand, accounts for the higher-order motion and appearance consistency between two temporally isolated tracklets. Our approach is modeled as a binary labeling problem and solved using the efficient Quadratic Pseudo-Boolean Optimization (QPBO). It yields promising tracking performance on the challenging PETS09 and MOT16 dataset. Our code will be made publicly available upon the acceptance of the manuscript.

Video super-resolution (SR) aims at estimating a high-resolution (HR) video sequence from a low-resolution (LR) one. Given that deep learning has been successfully applied to the task of single image SR, which demonstrates the strong capability of neural networks for modeling spatial relation within one single image, the key challenge to conduct video SR is how to efficiently and effectively exploit the temporal dependency among consecutive LR frames other than the spatial relation. However, this remains challenging because complex motion is difficult to model and can bring detrimental effects if not handled properly. We tackle the problem of learning temporal dynamics from two aspects. First, we propose a temporal adaptive neural network that can adaptively determine the optimal scale of temporal dependency. Inspired by the Inception module in GoogLeNet [1], filters of various temporal scales are applied to the input LR sequence before their responses are adaptively aggregated, in order to fully exploit the temporal relation among consecutive LR frames. Second, we decrease the complexity of motion among neighboring frames using a spatial alignment network that can be end-to-end trained with the temporal adaptive network and has the merit of increasing the robustness to complex motion and the efficiency compared to competing image alignment methods. We provide a comprehensive evaluation of the temporal adaptation and the spatial alignment modules. We show the temporal adaptive design considerably improve SR quality over its plain counterparts, and the spatial alignment network is able to attain comparable SR performance with the sophisticated optical flow based approach, but requires much less running time. Overall our proposed model with learned temporal dynamics is shown to achieve state-of-the-art SR results in terms of not only spatial consistency but also temporal coherence on public video datasets. More information can be found in.

Emotion recognition from facial expressions is tremendously useful, especially when coupled with smart devices and wireless multimedia applications. However, the inadequate network bandwidth often limits the spatial resolution of the transmitted video, which will heavily degrade the recognition reliability. We develop a novel framework to achieve robust emotion recognition from low bit rate video. While video frames are downsampled at the encoder side, the decoder is embedded with a deep network model for joint super-resolution (SR) and recognition. Notably, we propose a novel max-mix training strategy, leading to a single “One-for-All” model that is remarkably robust to a vast range of downsampling factors. That makes our framework well adapted for the varied bandwidths in real transmission scenarios, without hampering scalability or efficiency. The proposed framework is evaluated on the AVEC 2016 benchmark, and demonstrates significantly improved stand-alone recognition performance, as well as rate-distortion (R-D) performance, than either directly recognizing from LR frames, or separating SR and recognition.

Image matting is a fundamental computer vision problem and has many applications. Previous algorithms have poor performance when an image has similar foreground and background colors or complicated textures. The main reasons are prior methods 1) only use low-level features and 2) lack high-level context. In this paper, we propose a novel deep learning based algorithm that can tackle both these problems. Our deep model has two parts. The first part is a deep convolutional encoder-decoder network that takes an image and the corresponding trimap as inputs and predict the alpha matte of the image. The second part is a small convolutional network that refines the alpha matte predictions of the first network to have more accurate alpha values and sharper edges. In addition, we also create a large-scale image matting dataset including 49300 training images and 1000 testing images. We evaluate our algorithm on the image matting benchmark, our testing set, and a wide variety of real images. Experimental results clearly demonstrate the superiority of our algorithm over previous methods.

Video super-resolution (SR) aims to generate a high-resolution (HR) frame from multiple low-resolution (LR) frames in a local temporal window. The inter-frame temporal relation is as crucial as the intra-frame spatial relation for tackling this problem. However, how to utilize temporal information efficiently and effectively remains challenging since complex motion is difficult to model and can introduce adverse effects if not handled properly. We address this problem from two aspects. First, we propose a temporal adaptive neural network that can adaptively determine the optimal scale of temporal dependency. Filters on various temporal scales are applied to the input LR sequence before their responses are adaptively aggregated. Second, we reduce the complexity of motion between neighboring frames using a spatial alignment network which is much more robust and efficient than competing alignment methods and can be jointly trained with the temporal adaptive network in an end-to-end manner. Our proposed models with learned temporal dynamics are systematically evaluated on public video datasets and achieve state-of-the-art SR results compared with other recent video SR approaches. Both of the temporal adaptation and the spatial alignment modules are demonstrated to considerably improve SR quality over their plain counterparts.

The practical application of Computed Tomography (CT) faces the dilemma between higher image resolution and less X-ray exposure for patients, motivating the research on CT super-resolution (SR). In this paper, we apply state-of-the-art SR techniques to reconstruct CT images using two proposed advanced CT SR models based on Convolutional Neural Networks (CNNs) and residual learning: a single-slice CT SR network (S-CTSRN), and a multi-slice CT SR network (M-CTSRN). S-CTSRN improves the high-frequency feature extraction by incorporating the residual learning strategy, while M-CTSRN further utilizes the coherence between neighboring CT slices for better SR reconstruction. We evaluate both models on a large-scale CT dataset1, and obtain competitive results both quantitatively and qualitatively.

We propose the doubly sparsifying network (DSN), by drawing inspirations from the double sparsity model for dictionary learning. DSN emphasizes the joint utilization of both the problem structure and the parameter structure. It simultaneously sparsifies the output features and the learned model parameters, under one unified framework. DSN enjoys intuitive model interpretation, compact model size and low complexity. We compare DSN against a few carefully-designed baselines, to verify its consistently superior performance in a wide range of settings. Encouraged by its robustness to insufficient training data, we explore the applicability of DSN in brain signal processing that has been a challenging interdisciplinary area. DSN is evaluated for two mainstream tasks, electroencephalographic (EEG) signal classification and blood oxygenation level dependent (BOLD) response prediction, both achieving promising results.

Image aesthetics assessment has been challenging due to its subjective nature. Inspired by the Chatterjee's visual neuroscience model, we design Deep Chatterjee's Machine (DCM) tailored for this task. DCM first learns attributes through the parallel supervised pathways, on a variety of selected feature dimensions. A high-level synthesis network is trained to associate and transform those attributes into the overall aesthetics rating. We then extend DCM to predicting the distribution of human ratings, since aesthetics ratings are often subjective. We also highlight our first-of-its-kind study of label-preserving transformations in the context of aesthetics assessment, which leads to an effective data augmentation approach. Experimental results on the AVA dataset show that DCM gains significant performance improvement, compared to other state-of-the-art models.

Learning with recurrent neural networks (RNNs) on long sequences is a notoriously difficult task. There are three major challenges: 1) complex dependencies, 2) vanishing and exploding gradients, and 3) efficient parallelization. In this paper, we introduce a simple yet effective RNN connection structure, the DilatedRNN, which simultaneously tackles all of these challenges. The proposed architecture is characterized by multi-resolution dilated recurrent skip connections and can be combined flexibly with diverse RNN cells. Moreover, the DilatedRNN reduces the number of parameters needed and enhances training efficiency significantly, while matching state-of-the-art performance (even with standard RNN cells) in tasks involving very long-term dependencies. To provide a theory-based quantification of the architecture's advantages, we introduce a memory capacity measure, the mean recurrent length, which is more suitable for RNNs with long skip connections than existing measures. We rigorously prove the advantages of the DilatedRNN over other recurrent neural architectures. The code for our method is publicly available at https://github.com/code-terminator/DilatedRNN.

The increasing popularity of real-world recommender systems produces data continuously and rapidly, and it becomes more realistic to study recommender systems under streaming scenarios. Data streams present distinct properties such as temporally ordered, continuous and high-velocity, which poses tremendous challenges to traditional recommender systems. In this paper, we investigate the problem of recommendation with stream inputs. In particular, we provide a principled framework termed sRec, which provides explicit continuous-time random process models of the creation of users and topics, and of the evolution of their interests. A variational Bayesian approach called recursive meanfield approximation is proposed, which permits computationally efficient instantaneous on-line inference. Experimental results on several real-world datasets demonstrate the advantages of our sRec over other state-of-the-arts.

It is challenging to find relevant data for research and development purposes in the geospatial big data era. One long-standing problem in data discovery is locating, assimilating and utilizing the semantic context for a given query. Most research in the geospatial domain has approached this problem in one of two ways: building a domain-specific ontology manually or discovering automatically, semantic relationships using metadata and machine learning techniques. The former relies on rich expert knowledge but is static, costly and labor intensive, whereas the second is automatic and prone to noise. An emerging trend in information science takes advantage of large-scale user search histories, which are dynamic but subject to user- and crawler-generated noise. Leveraging the benefits of these three approaches and avoiding their weaknesses, a novel methodology is proposed to (1) discover vocabulary-based semantic relationships from user search histories and clickstreams, (2) refine the similarity calculation methods from existing ontologies and (3) integrate the results of ontology, metadata, user search history and clickstream analysis to better determine their semantic relationships. An accuracy assessment by domain experts for the similarity values indicates an 83% overall accuracy for the top 10 related terms over randomly selected sample queries. This research functions as an example for building vocabulary-based semantic relationships for different geographical domains to improve various aspects of data discovery, including the accuracy of the vocabulary relationships of commonly used search terms.

In this paper, we present a label transfer model from texts to images for image classification tasks. The problem of image classification is often much more challenging than text classification. On one hand, labeled text data is more widely available than the labeled images for classification tasks. On the other hand, text data tends to have natural semantic interpretability, and they are often more directly related to class labels. On the contrary, the image features are not directly related to concepts inherent in class labels. One of our goals in this paper is to develop a model for revealing the functional relationships between text and image features as to directly transfer intermodal and intramodal labels to annotate the images. This is implemented by learning a transfer function as a bridge to propagate the labels between two multimodal spaces. However, the intermodal label transfers could be undermined by blindly transferring the labels of noisy texts to annotate images. To mitigate this problem, we present an intramodal label transfer process, which complements the intermodal label transfer by transferring the image labels instead when relevant text is absent from the source corpus. In addition, we generalize the inter-modal label transfer to zero-shot learning scenario where there are only text examples available to label unseen classes of images without any positive image examples. We evaluate our algorithm on an image classification task and show the effectiveness with respect to the other compared algorithms.

Unsupervised outlier detection is a vital task and has high impact on a wide variety of applications domains, such as image analysis and video surveillance. It also gains long-standing attentions and has been extensively studied in multiple research areas. Detecting and taking action on outliers as quickly as possible are imperative in order to protect network and related stakeholders or to maintain the reliability of critical systems. However, outlier detection is difficult due to the one class nature and challenges in feature construction. Sequential anomaly detection is even harder with more challenges from temporal correlation in data, as well as the presence of noise and high dimensionality. In this paper, we introduce a novel deep structured framework to solve the challenging sequential outlier detection problem. We use autoencoder models to capture the intrinsic difference between outliers and normal instances and integrate the models to recurrent neural networks that allow the learning to make use of previous context as well as make the learners more robust to warp along the time axis. Furthermore, we propose to use a layerwise training procedure, which significantly simplifies the training procedure and hence helps achieve efficient and scalable training. In addition, we investigate a fine-tuning step to update all parameters set by incorporating the temporal correlation in the sequence. We further apply our proposed models to conduct systematic experiments on five real-world benchmark data sets. Experimental results demonstrate the effectiveness of our model, compared with other state-of-the-art approaches.

Recent advances in multimedia ontology have resulted in a number of concept models, e.g., large-scale concept for multimedia and Mediamill 101, which are accessible and public to other researchers. However, most current research effort still focuses on building new concepts from scratch, very few work explores the appropriate method to construct new concepts upon the existing models already in the warehouse. To address this issue, we propose a new framework in this paper, termed LEarning Structured Model by Probabilistic loGic Ontology Tree for MultiM edia (LEGO1-MM), which can seamlessly integrate both the new target training examples and the existing primitive concept models to infer the more complex concept models. LEGO-MM treats the primitive concept models as the lego toy to potentially construct an unlimited vocabulary of new concepts. Specifically, we first formulate the logic operations to be the lego connectors to combine the existing concept models hierarchically in probabilistic logic ontology trees. Then, we incorporate new target training information simultaneously to efficiently disambiguate the underlying logic tree and correct the error propagation. Extensive experiments are conducted on a large vehicle domain data set from ImageNet. The results demonstrate that LEGO-MM has significantly superior performance over the existing state-of-the-art methods, which build new concept models from scratch.

Minimum-cost flow algorithms have recently achieved state-of-the-art results in multi-object tracking. However, they rely on the whole image sequence as input. When deployed in real-time applications or in distributed settings, these algorithms first operate on short batches of frames and then stitch the results into full trajectories. This decoupled strategy is prone to errors because the batch-based tracking errors may propagate to the final trajectories and cannot be corrected by other batches. In this paper, we propose a greedy batch-based minimum-cost flow approach for tracking multiple objects. Unlike existing approaches that conduct batch-based tracking and stitching sequentially, we optimize consecutive batches jointly so that the tracking results on one batch may benefit the results on the other. Specifically, we apply a generalized minimum-cost flows (MCF) algorithm on each batch and generate a set of conflicting trajectories. These trajectories comprise the ones with high probabilities, but also those with low probabilities potentially missed by detectors and trackers. We then apply the generalized MCF again to obtain the optimal matching between trajectories from consecutive batches. Our proposed approach is simple, effective, and does not require training. We demonstrate the power of our approach on data sets of different scenarios.

Despite its nonconvex nature, ℓ0 sparse approximation is desirable in many theoretical and application cases. We study the ℓ0 sparse approximation problem with the tool of deep learning, by proposing Deep ℓ0 Encoders. Two typical forms, the ℓ0 regularized problem and the M-sparse problem, are investigated. Based on solid iterative algorithms, we model them as feed-forward neural networks, through introducing novel neurons and pooling functions. Enforcing such structural priors acts as an effective network regularization. The deep encoders also enjoy faster inference, larger learning capacity, and better scalability compared to conventional sparse coding solutions. Furthermore, under task-driven losses, the models can be conveniently optimized from end to end. Numerical results demonstrate the impressive performances of the proposed encoders.

We propose Epitomic Image Super-Resolution (ESR) to enhance the current internal SR methods that exploit the self-similarities in the input. Instead of local nearest neighbor patch matching used in most existing internal SR methods, ESR employs epitomic patch matching that features robustness to noise, and both local and non-local patch matching. Extensive objective and subjective evaluation demonstrate the effectiveness and advantage of ESR on various images.

Visual recognition research often assumes a sufficient resolution of the region of interest (ROI). That is usually violated in practice, inspiring us to explore the Very Low Resolution Recognition (VLRR) problem. Typically, the ROI in a VLRR problem can be smaller than 16 16 pixels, and is challenging to be recognized even by human experts. We attempt to solve the VLRR problem using deep learning methods. Taking advantage of techniques primarily in super resolution, domain adaptation and robust regression, we formulate a dedicated deep learning method and demonstrate how these techniques are incorporated step by step. Any extra complexity, when introduced, is fully justified by both analysis and simulation results. The resulting Robust Partially Coupled Networks achieves feature enhancement and recognition simultaneously. It allows for both the flexibility to combat the LR-HR domain mismatch, and the robustness to outliers. Finally, the effectiveness of the proposed models is evaluated on three different VLRR tasks, including face identification, digit recognition and font recognition, all of which obtain very impressive performances.

In this paper, we design a Deep Dual-Domain (D3) based fast restoration model to remove artifacts of JPEG compressed images. It leverages the large learning capacity of deep networks, as well as the problem-specific expertise that was hardly incorporated in the past design of deep architectures. For the latter, we take into consideration both the prior knowledge of the JPEG compression scheme, and the successful practice of the sparsity-based dual-domain approach. We further design the One-Step Sparse Inference (1-SI) module, as an efficient and lightweighted feed-forward approximation of sparse coding. Extensive experiments verify the superiority of the proposed D3 model over several state-of-the-art methods. Specifically, our best model is capable of outperforming the latest deep model for around 1 dB in PSNR, and is 30 times faster.

Interactive object selection is a very important research problem and has many applications. Previous algorithms require substantial user interactions to estimate the foreground and background distributions. In this paper, we present a novel deep-learning-based algorithm which has much better understanding of objectness and can reduce user interactions to just a few clicks. Our algorithm transforms user-provided positive and negative clicks into two Euclidean distance maps which are then concatenated with the RGB channels of images to compose (image, user interactions) pairs. We generate many of such pairs by combining several random sampling strategies to model users' click patterns and use them to finetune deep Fully Convolutional Networks (FCNs). Finally the output probability maps of our FCN-8s model is integrated with graph cut optimization to refine the boundary segments. Our model is trained on the PASCAL segmentation dataset and evaluated on other datasets with different object classes. Experimental results on both seen and unseen objects demonstrate that our algorithm has a good generalization ability and is superior to all existing interactive object selection approaches.

Clustering is an unsupervised learning approach that explores data and seeks groups of similar objects. Many classical clustering models such as k-means and DBSCAN are based on heuristics algorithms and suffer from local optimal solutions and numerical instability. Recently convex clustering has received increasing attentions, which leverages the sparsity inducing norms and enjoys many attractive theoretical properties. However, convex clustering is based on Euclidean distance and is thus not robust against outlier features. Since the outlier features are very common especially when dimensionality is high, the vulnerability has greatly limited the applicability of convex clustering to analyze many real-world datasets. In this paper, we address the challenge by proposing a novel robust convex clustering method that simultaneously performs convex clustering and identifies outlier features. Specifically, the proposed method learns to decompose the data matrix into a clustering structure component and a group sparse component that captures feature outliers. We develop a block coordinate descent algorithm which iteratively performs convex clustering after outliers features are identified and eliminated. We also propose an efficient algorithm for solving the convex clustering by exploiting the structures on its dual problem. Moreover, to further illustrate the statistical stability, we present the theoretical performance bound of the proposed clustering method. Empirical studies on synthetic data and real-world data demonstrate that the proposed robust convex clustering can detect feature outliers as well as improve cluster quality.

We consider the task of dimensional emotion recognition on video data using deep learning. While several previous methods have shown the benefits of training temporal neural network models such as recurrent neural networks (RNNs) on hand-crafted features, few works have considered combining convolutional neural networks (CNNs) with RNNs. In this work, we present a system that performs emotion recognition on video data using both CNNs and RNNs, and we also analyze how much each neural network component contributes to the system's overall performance. We present our findings on videos from the Audio/Visual+Emotion Challenge (AV+EC2015). In our experiments, we analyze the effects of several hyperparameters on overall performance while also achieving superior performance to the baseline and other competing methods.

Data of many problems in real-world systems such as link prediction and one-class recommendation share common characteristics. First, data are in the form of positive unlabeled (PU) measurements (e.g. Twitter "following", Facebook "like", etc.) that do not provide negative information, which can be naturally represented as networks. Second, in the era of big data, such data are generated temporally-ordered, continuously and rapidly, which determines its streaming nature. These common characteristics allow us to unify many problems into a novel framework -- PU learning in streaming networks. In this paper, a principled probabilistic approach SPU is proposed to leverage the characteristics of the streaming PU inputs. In particular, SPU captures temporal dynamics and provides real-time adaptations and predictions by identifying the potential negative signals concealed in unlabeled data. Our empirical results on various real-world datasets demonstrate the effectiveness of the proposed framework over other state-of-the-art methods in both link prediction and recommendation.

In present object detection systems, the deep convolutional neural networks (CNNs) are utilized to predict bounding boxes of object candidates, and have gained performance advantages over the traditional region proposal methods. However, existing deep CNN methods assume the object bounds to be four independent variables, which could be regressed by the l2 loss separately. Such an oversimplified assumption is contrary to the well-received observation, that those variables are correlated, resulting to less accurate localization. To address the issue, we firstly introduce a novel Intersection over Union (IoU) loss function for bounding box prediction, which regresses the four bounds of a predicted box as a whole unit. By taking the advantages of IoU loss and deep fully convolutional networks, the UnitBox is introduced, which performs accurate and efficient localization, shows robust to objects of varied shapes and scales, and converges fast. We apply UnitBox on face detection task and achieve the best performance among all published methods on the FDDB benchmark.

While sparse coding-based clustering methods have shown to be successful, their bottlenecks in both efficiency and scalability limit the practical usage. In recent years, deep learning has been proved to be a highly effective, efficient and scalable feature learning tool. In this paper, we propose to emulate the sparse coding-based clustering pipeline in the context of deep learning, leading to a carefully crafted deep model benefiting from both. A feed-forward network structure, named TAGnet, is constructed based on a graph-regularized sparse coding algorithm. It is then trained with task-specific loss functions from end to end. We discover that connecting deep learning to sparse coding benefits not only the model performance, but also its initialization and interpretation. Moreover, by introducing auxiliary clustering tasks to the intermediate feature hierarchy, we formulate DTAGnet and obtain a further performance boost. Extensive experiments demonstrate that the proposed model gains remarkable margins over several state-of-the-art methods.

Single image super-resolution (SR) is an ill-posed problem, which tries to recover a high-resolution image from its low-resolution observation. To regularize the solution of the problem, previous methods have focused on designing good priors for natural images, such as sparse representation, or directly learning the priors from a large data set with models, such as deep neural networks. In this paper, we argue that domain expertise from the conventional sparse coding model can be combined with the key ingredients of deep learning to achieve further improved results. We demonstrate that a sparse coding model particularly designed for SR can be incarnated as a neural network with the merit of end-to-end optimization over training data. The network has a cascaded structure, which boosts the SR performance for both fixed and incremental scaling factors. The proposed training and testing schemes can be extended for robust handling of images with additional degradation, such as noise and blurring. A subjective assessment is conducted and analyzed in order to thoroughly evaluate various SR techniques. Our proposed model is tested on a wide range of images, and it significantly outperforms the existing state-of-the-art methods for various scaling factors both quantitatively and perceptually.

Exploring image structure is a long-standing yet important research subject in the computer vision community. In this paper, we focus on understanding image structure inspired by the “simple-to-complex” biological evidence. A hierarchical shape parsing strategy is proposed to partition and organize image components into a hierarchical structure in the scale space. To improve the robustness and flexibility of image representation, we further bundle the image appearances into hierarchical parsing trees. Image descriptions are subsequently constructed by performing a structural pooling, facilitating efficient matching between the parsing trees. We leverage the proposed hierarchical shape parsing to study two exemplar applications including edge scale refinement and unsupervised “objectness” detection. We show competitive parsing performance comparing to the state-of-the-arts in above scenarios with far less proposals, which thus demonstrates the advantage of the proposed parsing scheme.

While feedforward deep convolutional neural networks (CNNs) have been a great success in computer vision, it is important to note that the human visual cortex generally contains more feedback than feedforward connections. In this paper, we will briefly introduce the background of feedbacks in the human visual cortex, which motivates us to develop a computational feedback mechanism in deep neural networks. In addition to the feedforward inference in traditional neural networks, a feedback loop is introduced to infer the activation status of hidden layer neurons according to the "goal" of the network, e.g., high-level semantic labels. We analogize this mechanism as "Look and Think Twice." The feedback networks help better visualize and understand how deep neural networks work, and capture visual attention on expected objects, even in images with cluttered background and multiple objects. Experiments on ImageNet dataset demonstrate its effectiveness in solving tasks such as image classification and object localization.

Deep learning techniques have been successfully applied in many areas of computer vision, including low-level image restoration problems. For image super-resolution, several models based on deep neural networks have been recently proposed and attained superior performance that overshadows all previous handcrafted models. The question then arises whether large-capacity and data-driven models have become the dominant solution to the ill-posed super-resolution problem. In this paper, we argue that domain expertise represented by the conventional sparse coding model is still valuable, and it can be combined with the key ingredients of deep learning to achieve further improved results. We show that a sparse coding model particularly designed for super-resolution can be incarnated as a neural network, and trained in a cascaded structure from end to end. The interpretation of the network based on sparse coding leads to much more efficient and effective training, as well as a reduced model size. Our model is evaluated on a wide range of images, and shows clear advantage over existing state-of-the-art methods in terms of both restoration accuracy and human subjective quality.

3D human body shape matching has large potential on many real world applications, especially with the recent advances in the 3D range sensing technology. We address this problem by proposing a novel holistic human body shape descriptor called BodyPrint. To compute the bodyprint for a given body scan, we fit a deformable human body mesh and project the mesh parameters to a low-dimensional subspace which improves discriminability across different persons. Experiments are carried out on three real-world human body datasets to demonstrate that BodyPrint is robust to pose variation as well as missing information and sensor noise. It improves the matching accuracy significantly compared to conventional 3D shape matching techniques using local features. To facilitate practical applications where the shape database may grow over time, we also extend our learning framework to handle online updates.

Data embedding is used in many machine learning applications to create low-dimensional feature representations, which preserves the structure of data points in their original space. In this paper, we examine the scenario of a heterogeneous network with nodes and content of various types. Such networks are notoriously difficult to mine because of the bewildering combination of heterogeneous contents and structures. The creation of a multidimensional embedding of such data opens the door to the use of a wide variety of off-the-shelf mining techniques for multidimensional data. Despite the importance of this problem, limited efforts have been made on embedding a network of scalable, dynamic and heterogeneous data. In such cases, both the content and linkage structure provide important cues for creating a unified feature representation of the underlying network. In this paper, we design a deep embedding algorithm for networked data. A highly nonlinear multi-layered embedding function is used to capture the complex interactions between the heterogeneous data in a network. Our goal is to create a multi-resolution deep embedding function, that reflects both the local and global network structures, and makes the resulting embedding useful for a variety of data mining tasks. In particular, we demonstrate that the rich content and linkage information in a heterogeneous network can be captured by such an approach, so that similarities among cross-modal data can be measured directly in a common embedding space. Once this goal has been achieved, a wide variety of data mining problems can be solved by applying off-the-shelf algorithms designed for handling vector representations. Our experiments on real-world network datasets show the effectiveness and scalability of the proposed algorithm as compared to the state-of-the-art embedding methods.

We develop the DeepFont system, a large-scale learning-based solution for automatic font identification, organization and selection. In this proposed technical demonstration, we will give our audience a tour to the DeepFont system, with the focus on its impacts on real consumer products, including but not limited to: 1) a cloud-based iOS App for font recognition; 2) a web-based tool for font similarity evaluation and discovery.

As font is one of the core design concepts, automatic font identification and similar font suggestion from an image or photo has been on the wish list of many designers. We study the Visual Font Recognition (VFR) problem [4] LFE, and advance the state-of-the-art remarkably by developing the DeepFont system. First of all, we build up the first available large-scale VFR dataset, named AdobeVFR, consisting of both labeled synthetic data and partially labeled real-world data. Next, to combat the domain mismatch between available training and testing data, we introduce a Convolutional Neural Network (CNN) decomposition approach, using a domain adaptation technique based on a Stacked Convolutional Auto-Encoder (SCAE) that exploits a large corpus of unlabeled real-world text images combined with synthetic data preprocessed in a specific way. Moreover, we study a novel learning-based model compression approach, in order to reduce the DeepFont model size without sacrificing its performance. The DeepFont system achieves an accuracy of higher than 80% (top-5) on our collected dataset, and also produces a good font similarity measure for font selection and suggestion. We also achieve around 6 times compression of the model without any visible loss of recognition accuracy.

Social streams have become very popular in recent years because of the increasing popularity of social media sites such as Twitter, and Facebook. Such social media sites create huge streams of data, which can be leveraged for a wide variety of applications. In this paper, we will focus on the classification problem for social streams. Unfortunately, such streams are extremely noisy, and contain large volumes of information, with information about network linkages between the participants exchanging messages. This is additional social information, associated with the text stream, which can be very helpful for classification. We combine an LSH method with an incremental SVM model in order to design an effective and efficient social context-sensitive streaming classifier for this scenario. The LSH model is used for learning the social context, and the SVM model is used for more effective classification within this context. We will present experimental results, which show the effectiveness of our techniques over a wide variety of other methods.

We study the complementary behaviors of external and internal examples in image restoration, and are motivated to formulate a composite dictionary design framework. The composite dictionary consists of the global part learned from external examples, and the sample-specific part learned from internal examples. The dictionary atoms in both parts are further adaptively weighted to emphasize their model statistics. Experiments demonstrate that the joint utilization of external and internal examples leads to substantial improvements, with successful applications in image denoising and super resolution.

Classifying large-scale image data into object categories is an important problem that has received increasing research attention. Given the huge amount of data, non-parametric approaches such as nearest neighbor classifiers have shown promising results, especially when they are underpinned by a learned distance or similarity measurement. Although metric learning has been well studied in the past decades, most existing algorithms are impractical to handle large-scale data sets. In this paper, we present an image similarity learning method that can scale well in both the number of images and the dimensionality of image descriptors. To this end, similarity comparison is restricted to each sample's local neighbors and a discriminative similarity measure is induced from large margin neighborhood embedding. We also exploit the ensemble of projections so that high-dimensional features can be processed in a set of lower-dimensional subspaces in parallel. The efficiency and scalability of our proposed model are validated on several data sets with scales varying from tens of thousands to one million images.

In this letter, we introduce a subcategory-aware object detection framework to detect generic object classes with high intra-class variance. Motivated by the observation that the object appearance demonstrates some clustering property, we split the training data into subcategories and train a detector for each subcategory. Since the proposed ensemble of detectors relies heavily on subcategory clustering, we propose an effective subcategories generation method that is tuned for the detection task. More specifically, we first initialize subcategories by constrained spectral clustering based on mid-level image features used in object recognition. Then we jointly learn the ensemble detectors and the latent subcategories in an alternative manner. Our performance on the PASCAL VOC 2007 detection challenges and INRIA Person dataset is comparable with state-of-the-art, even with much less computational cost.

Inspired by the recent image feature learning work, we propose a novel key point detection approach for object tracking. Our approach can select mid-level interest key points by max pooling over the local descriptor responses from a set of filters. Linear filters are first learned from targets in first frames. Then max pooling is performed over data driven spatial supporting field to detect discriminant key points, and thus the detected key points bear higher level semantic meanings, which we apply in tracking by structured key point matching. We show that our tracking system is robust to occlusions and cluttered background. Testing on several challenging tracking sequences, we demonstrate that our proposed tracking system can achieve competitive or better performances than the state-of-the-art trackers.

We present a semisupervised method for single-pixel classification of hyperspectral images. The proposed method is designed to address the special problematic characteristics of hyperspectral images, namely, high dimensionality of hyperspectral pixels, lack of labeled samples, and spatial variability of spectral signatures. To alleviate these problems, the proposed method features the following components. First, being a semisupervised approach, it exploits the wealth of unlabeled samples in the image by evaluating the confidence probability of the predicted labels, for each unlabeled sample. Second, we propose to jointly optimize the classifier parameters and the dictionary atoms by a task-driven formulation, to ensure that the learned features (sparse codes) are optimal for the trained classifier. Finally, it incorporates spatial information through adding a Laplacian smoothness regularization to the output of the classifier, rather than the sparse codes, making the spatial constraint more flexible. The proposed method is compared with a few comparable methods for classification of several popular data sets, and it produces significantly better classification results.

Single image super-resolution (SR) aims to estimate a high-resolution (HR) image from a low-resolution (LR) input. Image priors are commonly learned to regularize the, otherwise, seriously ill-posed SR problem, either using external LR-HR pairs or internal similar patterns. We propose joint SR to adaptively combine the advantages of both external and internal SR methods. We define two loss functions using sparse coding-based external examples, and epitomic matching based on internal examples, as well as a corresponding adaptive weight to automatically balance their contributions according to their reconstruction errors. Extensive SR results demonstrate the effectiveness of the proposed method over the existing state-of-the-art methods, and is also verified by our subjective evaluation study.

Link prediction is one of the most fundamental problems in graph modeling and mining. It has been studied in a wide range of scenarios, from uncovering missing links between different entities in databases, to recommending relations between people in social networks. In this problem, we wish to predict unseen links in a growing target network by exploiting existing structures in source networks. Most of the existing methods often assume that abundant links are available in the target network to build a model for link prediction. However, in many scenarios, the target network may be too sparse to enable robust inference process, which makes link prediction challenging with the paucity of link data. On the other hand, in many cases, other (more densely linked) auxiliary networks can be available that contains similar link structure relevant to that in the target network. The linkage information in the existing networks can be used in conjunction with the node attribute information in both networks in order to make more accurate link recommendations. Thus, this paper proposes the use of learning methods to perform link inference by transferring the link information from the source network to the target network. We also note that the source network may contain the link information irrelevant to the target network. This leads to cross-network bias between the networks, which makes the link model built upon the source network misaligned with the link structure of the target network. Therefore, we re-sample the source network to rectify such cross-network bias by maximizing the cross-network relevance measured by the node attributes, as well as preserving as rich link information as possible to avoid the loss of source link structure caused by the re-sampling algorithm. The link model based on the re-sampled source network can make more accurate link predictions on the target network with aligned link structures across the networks. We present experimental results illustrating the effectiveness of the approach.

L1-Graph has been proven to be effective in data clustering, which partitions the data space by using the sparse representation of the data as the similarity measure. However, the sparse representation is performed for each datum separately without taking into account the geometric structure of the data. Motivated by L1-Graph and manifold leaning, we propose Laplacian Regularized L1-Graph (LRℓ1-Graph) for data clustering. The sparse representations of LRℓ1-Graph are regularized by the geometric information of the data so that they vary smoothly along the geodesics of the data manifold by the graph Laplacian according to the manifold assumption. Moreover, we propose an iterative regularization scheme, where the sparse representation obtained from the previous iteration is used to build the graph Laplacian for the current iteration of regularization. The experimental results on real data sets demonstrate the superiority of our algorithm compared to L1-Graph and other competing clustering methods.

Image colorization adds color to grayscale images. It not only increases the visual appeal of grayscale images, but also enriches the information conveyed by scientific images that lack color information. We develop a new image colorization method, epitomic image colorization, which automatically transfers color from the reference color image to the target grayscale image by a robust feature matching scheme using a new feature representation, namely the heterogeneous feature epitome. As a generative model, heterogeneous feature epitome is a condensed representation of image appearance which is employed for measuring the dissimilarity between reference patches and target patches in a way robust to noise in the reference image. We build a Markov Random Field (MRF) model with the learned heterogeneous feature epitome from the reference image, and inference in the MRF model achieves robust feature matching for transferring color. Our method renders better colorization results than the current state-of-the-art automatic colorization methods in our experiments.

Exemplar-based clustering methods partition the data space and identify the representative, or the exemplar, of each cluster. With the number of clusters adaptively determined, exemplar-based clustering methods are appealing since they avoid or alleviate the difficult task of estimating the latent parameters in case of complex models and high dimensionality of the data. Most exemplar-based clustering methods are based on generative models, where the exemplars serve as the parameters of the generative models. However, generative models do not consider the discriminative capability of the cluster boundaries explicitly described in discriminative models. In this paper, we present Discriminative Exemplar Clustering (DEC), that improves the discriminative power of exemplar-based clustering method by minimizing the misclassification error of the nonparametric unsupervised plug-in classifier while maintaining the appealing property of exemplar-based clustering. The optimization of DEC is performed in a pairwise Markov Random Field. Experimental results on synthetic and real data demonstrate the effectiveness of our method compared to other exemplar-based clustering methods.

Recent advances in distance function learning have demonstrated that learning a good distance metric can greatly improve the performance in a wide variety of tasks in data mining and web search. A major problem in such scenarios is the limited labeled knowledge available for learning the user intentions. Furthermore, distances are inherently local, where a single global distance function may not capture the distance structure well. A challenge here is that local distance learning is even harder when the labeled information available is limited, because the distance function varies with data locality. To address these issues, we propose a local metric learning algorithm termed Local Semantic Sensing (LSS), which augments the small amount of labeled data with unlabeled data in order to learn the semantic information in the manifold structure, and then integrated with supervised intentional knowledge in a local way. We present results in a retrieval application, which show that the approach significantly outperforms other state-of-the-art methods in the literature.

The problem of similarity learning is relevant to many data mining applications, such as recommender systems, classification, and retrieval. This problem is particularly challenging in the context of networks, which contain different aspects such as the topological structure, content, and user supervision. These different aspects need to be combined effectively, in order to create a holistic similarity function. In particular, while most similarity learning methods in networks such as Sim Rank utilize the topological structure, the user supervision and content are rarely considered. In this paper, a Factorized Similarity Learning (FSL) is proposed to integrate the link, node content, and user supervision into an uniform framework. This is learned by using matrix factorization, and the final similarities are approximated by the span of low rank matrices. The proposed framework is further extended to a noise-tolerant version by adopting a hinge-loss alternatively. To facilitate efficient computation on large scale data, a parallel extension is developed. Experiments are conducted on the DBLP and CoRA datasets. The results show that FSL is robust, efficient, and outperforms the state-of-the-art.

In this paper, we propose a novel saliency-based algorithm to detect foreground regions in highly dynamic scenes. We first convert input video frames to multiple patch-based feature maps. Then, we apply temporal saliency analysis to the pixels of each feature map. For each temporal set of co-located pixels, the feature distance of a point from its kthnearest neighbor is used to compute the temporal saliency. By computing and combining temporal saliency maps of different features, we obtain foreground likelihood maps. A simple segmentation method based on adaptive thresholding is applied to detect the foreground objects. We test our algorithm on images sequences of dynamic scenes, including public datasets and a new challenging wildlife dataset we constructed. The experimental results demonstrate the proposed algorithm achieves state-of-the-art results.

Bag-of-Feature (BoF) representations and spatial constraints have been popular in image classification research. One of the most successful methods uses sparse coding and spatial pooling to build discriminative features. However, minimizing the reconstruction error by sparse coding only considers the similarity between the input and codebooks. In contrast, this paper describes a novel feature learning approach for image classification by considering the dissimilarity between inputs and prototype images, or what we called reference basis (RB). First, we learn the feature representation by max-margin criterion between the input and the RB. The learned hyperplane is stored as the relative feature. Second, we propose an adaptive pooling technique to assemble multiple relative features generated by different RBs under the SVM framework, where the classifier and the pooling weights are jointly learned. Experiments based on three challenging datasets: Caltech-101, Scene 15 and Willow-Actions, demonstrate the effectiveness and generality of our framework.

Existing example-based super resolution (SR) methods are built upon either external-examples or self-examples. Although effective in certain cases, both methods suffer from their inherent limitation. This paper goes beyond these two classes of most common example-based SR approaches, and proposes a novel joint SR perspective. The joint SR exploits and maximizes the complementary advantages of external- and self-example based methods. We elaborate on exploitable priors for image components of different nature, and formulate their corresponding loss functions mathematically. Equipped with that, we construct a unified SR formulation, and propose an iterative joint super resolution (IJSR) algorithm to solve the optimization. Such a joint perspective approach leads to an impressive improvement of SR results both quantitatively and qualitatively.

The problem of image search has been studied extensively in recent years because of the large and increasing repositories of images on the web, social media, and other linked networks. Most of the available techniques for keyword-based image search on the web use the text in the surrounding or linked text in order to retrieve related images. Many image repositories on the web are built upon social media platforms such as Flickr. Such platforms provide a rich level of information in terms of the user linkage information to images, tags or other comments which are contributed by the users. It is reasonable to assume that the content of the images, users and other social cues such as tags and comments are often related to one another. Therefore, such cues can be useful for improving the effectiveness of search and ranking algorithms. In this paper, we propose SocialRank, which is a technique for using social hints in order to improve the image search and ranking process. Furthermore, we propose a holistic framework to combine social tags, social network text, linkage between actors and images, as well as the actual image features in order to create a ranking technique for image search. We design a PageRank-like method which can combine these different methods in order to provide an effective method for image search and ranking in social networks.

In this paper, a novel discriminative dictionary learning method is proposed for sparse-representation-based classification (SRC) to label highly dimensional hyperspectral imagery (HSI). In SRC, a dictionary is conventionally constructed using all of the training pixels, which is not only inefficient due to the large size of typical HSI images but also ineffective in capturing class-discriminative information crucial for classification. We address the dictionary design problem with the inspiration from the learning vector quantization technique and propose a hinge loss function that is directly related to the classification task as the objective function for dictionary learning. The resulting online learning procedure systematically “pulls” and “pushes” dictionary atoms so that they become better adapted to distinguish between different classes. In addition, the spatial context for a test pixel within its local neighborhood is modeled using a Bayesian graph model and is incorporated with the sparse representation of a single test pixel in a unified probabilistic framework, which enables further refinement of our dictionary to capture the spatial class dependence that complements the spectral information. Experiments on different HSI images demonstrate that the dictionaries optimized using our method can achieve higher classification accuracy with substantially reduced dictionary size than using the whole training set. The proposed method also outperforms existing dictionary learning methods and attains the state-of-the-art results in both the spectral-only and spatial-spectral settings.

The goal of multilabel classification is to reveal the underlying label correlations to boost the accuracy of classification tasks. Most of the existing multilabel classifiers attempt to exhaustively explore dependency between correlated labels. It increases the risk of involving unnecessary label dependencies, which are detrimental to classification performance. Actually, not all the label correlations are indispensable to multilabel model. Negligible or fragile label correlations cannot be generalized well to the testing data, especially if there exists label correlation discrepancy between training and testing sets. To minimize such negative effect in the multilabel model, we propose to learn a sparse structure of label dependency. The underlying philosophy is that as long as the multilabel dependency cannot be well explained, the principle of parsimony should be applied to the modeling process of the label correlations. The obtained sparse label dependency structure discards the outlying correlations between labels, which makes the learned model more generalizable to future samples. Experiments on real world data sets show the competitive results compared with existing algorithms.

This paper considers the person verification problem in modern surveillance and video retrieval systems. The problem is to identify whether a pair of face or human body images is about the same person, even if the person is not seen before. Traditional methods usually look for a distance (or similarity) measure between images (e.g., by metric learning algorithms), and make decisions based on a fixed threshold. We show that this is nevertheless insufficient and sub-optimal for the verification problem. This paper proposes to learn a decision function for verification that can be viewed as a joint model of a distance metric and a locally adaptive thresholding rule. We further formulate the inference on our decision function as a second-order large-margin regularization problem, and provide an efficient algorithm in its dual from. We evaluate our algorithm on both human body verification and face verification problems. Our method outperforms not only the classical metric learning algorithm including LMNN and ITML, but also the state-of-the-art in the computer vision community.

Sparse representation provides an efficient description for high-dimensional Hyperspectral Imagery (HSI) and also encodes discriminative information useful for classification. However, due to the large size of typical HSI images, the naive way to construct a dictionary with all training pixels is neither efficient nor practical. In this paper, a novel approach is proposed to design compact dictionary for Sparse Representation-based Classification (SRC). Inspired by Learning Vector Quantization (LVQ) techniques, we use a hinge loss function directly related to classification task as our objective function, and optimize the dictionary by exploiting the differentiable parts of sparse codes. The resultant dictionary updating procedure adapts the “push” and “pull” actions in LVQ to SRC, which is therefore named as Learning Sparse Representation-based Classification (LSRC). Experiments on different HSI images demonstrate that our LSRC approach can achieve higher classification accuracy with substantially smaller dictionary size than using the whole training set, and also outperforms existing dictionary learning methods.

Sparse Representation-based Classification (SRC) is a powerful tool in distinguishing signal categories which lie on different subspaces. Despite its wide application to visual recognition tasks, current understanding of SRC is solely based on a reconstructive perspective, which neither offers any guarantee on its classification performance nor provides any insight on how to design a discriminative dictionary for SRC. In this paper, we present a novel perspective towards SRC and interpret it as a margin classifier. The decision boundary and margin of SRC are analyzed in local regions where the support of sparse code is stable. Based on the derived margin, we propose a hinge loss function as the gauge for the classification performance of SRC. A stochastic gradient descent algorithm is implemented to maximize the margin of SRC and obtain more discriminative dictionaries. Experiments validate the effectiveness of the proposed approach in predicting classification performance and improving dictionary quality over reconstructive ones. Classification results competitive with other state-of-the-art sparse coding methods are reported on several data sets.

The problem of link inference has been widely studied in a variety of social networking scenarios. In this problem, we wish to predict future links in a growing network with the use of the existing network structure. However, most of the existing methods work well only if a significant number of links are already available in the network for the inference process. In many scenarios, the existing network may be too sparse, and may have too few links to enable meaningful learning mechanisms. This paucity of linkage information can be challenging for the link inference problem. However, in many cases, other (more densely linked) networks may be available which show similar linkage structure in terms of underlying attribute information in the nodes. The linkage information in the existing networks can be used in conjunction with the node attribute information in both networks in order to make meaningful link recommendations. Thus, this paper introduces the use of transfer learning methods for performing cross-network link inference. We present experimental results illustrating the effectiveness of the approach.

Recent advances in Multimedia research have generated a large collection of concept models, e.g., LSCOM and Media mill 101, which become accessible to other researchers. While most current research effort still focuses on building new concepts from scratch, little effort has been made on constructing new concepts upon the existing models already in the warehouse. To address this issue, we develop a new framework in this paper, termed LEGO, to seamlessly integrate both the new target training examples and the existing primitive concept models. LEGO treats the primitive concept models as a lego toy to potentially construct an unlimited vocabulary of new concepts. Specifically, LEGO first formulates the logic operations to be the lego connectors to combine existing concept models hierarchically in probabilistic logic ontology trees. LEGO then simultaneously incorporates new target training information to efficiently disambiguate the underlying logic tree and correct the error propagation. We present extensive experimental results on a large vehicle domain data set from Image Net, and demonstrate significantly superior performance over existing state-of-the-art approaches which build new concept models from scratch.

An effective and efficient image contour detector is highly desired due to its wide applications in computer vision and multimedia retrieval. However, the state-of-the-art image contour detection algorithms are very computationally intensive, and thus impractical for web-scale applications. In this work, we study the relationship between edge detection and contour detection, based on which an edge-based image contour detection algorithm is proposed. This algorithm fully makes use of cheap edge information for efficiency purpose. The experiments on benchmark data sets show that, the proposed contour detector works much faster than existing state-of-the-art algorithms while maintaining high accuracy, and thus suitable for large-scale applications.

A novel problem of object recognition with dynamically allocated sensing resources is considered in this paper. We call this problem opportunistic sensing since prior knowledge about the correlation between class label and signal distribution is exploited as early as in data acquisition. Two forms of sensing parameters — discrete sensor index and continuous linear measurement vector — are optimized within the same maximum negative entropy framework. The computationally intractable expected entropy is approximated using unscented transform for Gaussian models, and we solve the problem using a gradient-based method. Our formulation is theoretically shown to be closely related to the maximum mutual information criterion for sensor selection and linear feature extraction techniques such as PCA, LDA, and CCA. The proposed approach is validated on multi-view vehicle classification and face recognition datasets, and remarkable improvement over baseline methods is demonstrated in the experiments.

Pair wise clustering methods, including the popular graph cut based approaches such as normalized cut, partition the data space into clusters by the pair wise affinity between data points. The success of pair wise clustering largely depends on the pair wise affinity function defined over data points coming from different clusters. Interpreting the pair wise affinity in a probabilistic framework, we build the relationship between pair wise clustering and unsupervised classification by learning the soft Nearest Neighbor (NN) classifier from unlabeled data, and search for the optimal partition of the data points by minimizing the generalization error of the learned classifier associated with the data partitions. Modeling the underlying distribution of the data by non-parametric kernel density estimation, the asymptotic generalization error of the unsupervised soft NN classification involves only the pair wise affinity between data points. Moreover, such error rate reduces to the well-known kernel form of graph cut in case of uniform data distribution, which provides another understanding of the kernel similarity used in Laplacian Eigenmaps [1] which also assumes uniform distribution. By minimizing the generalization error bound, we propose a new clustering algorithm. Our algorithm efficiently partition the data by inference in a pair wise MRF model. Experimental results demonstrate the effectiveness of our method.

The proliferation of location and GPS data streams which are collected in a wide variety of participatory sensing applications has created numerous possibilities for analysis of the underlying patterns of activity. Typically, the spatio-temporal patterns arising from such activity can be analyzed in order to determine the latent community structure in the underlying data. In this paper, we will examine the problem of online community detection from the location data collected from such social sensing applications in real time. Such data brings numerous challenges associated with it, in that they can be of a relatively large scale, and can be extremely noisy from the perspective of both data representation and analysis. Furthermore, the community structure in the underlying data cannot be directly inferred from the shape of the underlying trajectories, since a considerable amount of variation may exist in terms of trajectories of individuals belonging to the same community. In this paper, we will design online algorithms for community detection in social sensing applications. Our algorithm uses a robust and efficiently updateable model with the use of Gibbs sampling, and we will show its effectiveness and efficiency for social sensing applications.

Collective intelligence, which aggregates the shared information from large crowds, is often negatively impacted by unreliable information sources with the low quality data. This becomes a barrier to the effective use of collective intelligence in a variety of applications. In order to address this issue, we propose a probabilistic model to jointly assess the reliability of sources and find the true data. We observe that different sources are often not independent of each other. Instead, sources are prone to be mutually influenced, which makes them dependent when sharing information with each other. High dependency between sources makes collective intelligence vulnerable to the overuse of redundant (and possibly incorrect) information from the dependent sources. Thus, we reveal the latent group structure among dependent sources, and aggregate the information at the group level rather than from individual sources directly. This can prevent the collective intelligence from being inappropriately dominated by dependent sources. We will also explicitly reveal the reliability of groups, and minimize the negative impacts of unreliable groups. Experimental results on real-world data sets show the effectiveness of the proposed approach with respect to existing algorithms.

Browsing large audio archives is challenging because of the limitations of human audition and attention. However, this task becomes easier with a suitable visualization of the audio signal, such as a spectrogram transformed to make unusual audio events salient. This transformation maximizes the mutual information between an isolated event's spectrogram and an estimate of how salient the event appears in its surrounding context. When such spectrograms are computed and displayed with fluid zooming over many temporal orders of magnitude, sparse events in long audio recordings can be detected more quickly and more easily. In particular, in a 1/10-real-time acoustic event detection task, subjects who were shown saliency-maximized rather than conventional spectrograms performed significantly better. Saliency maximization also improves the mutual information between the ground truth of nonbackground sounds and visual saliency, more than other common enhancements to visualization.

A nonlocal kernel regression (NL-KR) model is presented in this paper for various image and video restoration tasks. The proposed method exploits both the nonlocal self-similarity and local structural regularity properties in natural images. The nonlocal self-similarity is based on the observation that image patches tend to repeat themselves in natural images and videos, and the local structural regularity observes that image patches have regular structures where accurate estimation of pixel values via regression is possible. By unifying both properties explicitly, the proposed NL-KR framework is more robust in image estimation, and the algorithm is applicable to various image and video restoration tasks. In this paper, we apply the proposed model to image and video denoising, deblurring, and superresolution reconstruction. Extensive experimental results on both single images and realistic video sequences demonstrate that the proposed framework performs favorably with previous works both qualitatively and quantitatively.

Exemplar-based clustering methods have been extensively shown to be effective in many clustering problems. They adaptively determine the number of clusters and hold the appealing advantage of not requiring the estimation of latent parameters, which is otherwise difficult in case of complicated parametric model and high dimensionality of the data. However, modeling arbitrary underlying distribution of the data is still difficult for existing exemplar-based clustering methods. We present Pairwise Exemplar Clustering (PEC) to alleviate this problem by modeling the underlying cluster distributions more accurately with non-parametric kernel density estimation. Interpreting the clusters as classes from a supervised learning perspective, we search for an optimal partition of the data that balances two quantities: 1 the misclassification rate of the data partition for separating the clusters; 2 the sum of within-cluster dissimilarities for controlling the cluster size. The broadly used kernel form of cut turns out to be a special case of our formulation. Moreover, we optimize the corresponding objective function by a new efficient algorithm for message computation in a pairwise MRF. Experimental results on synthetic and real data demonstrate the effectiveness of our method.

Most image understanding algorithms begin with the extraction of information thought to be relevant to the particular task. This is commonly known as feature extraction and has, up to this date, been a largely manual process, where a reasonable method is chosen through validation on the experimented dataset. In this work we propose a data driven, local histogram based feature extraction method that reduces the manual intervention during the feature computation process and improves on the performance of widely used gradient histogram based features (e.g., HOG). We demonstrate favorable object detection results against HOG on the Inria Pedestrian[7], Pascal 2007[10] data.

This paper introduces a probabilistic graphical model for continuous action recognition with two novel components: substructure transition model and discriminative boundary model. The first component encodes the sparse and global temporal transition prior between action primitives in state-space model to handle the large spatial-temporal variations within an action class. The second component enforces the action duration constraint in a discriminative way to locate the transition boundaries between actions more accurately. The two components are integrated into a unified graphical structure to enable effective training and inference. Our comprehensive experimental results on both public and in-house datasets show that, with the capability to incorporate additional information that had not been explicitly or efficiently modeled by previous methods, our proposed algorithm achieved significantly improved performance for continuous action recognition.

In this paper, we propose a bilevel sparse coding model for coupled feature spaces, where we aim to learn dictionaries for sparse modeling in both spaces while enforcing some desired relationships between the two signal spaces. We first present our new general sparse coding model that relates signals from the two spaces by their sparse representations and the corresponding dictionaries. The learning algorithm is formulated as a generic bilevel optimization problem, which is solved by a projected first-order stochastic gradient descent algorithm. This general sparse coding model can be applied to many specific applications involving coupled feature spaces in computer vision and signal processing. In this work, we tailor our general model to learning dictionaries for compressive sensing recovery and single image super-resolution to demonstrate its effectiveness. In both cases, the new sparse coding model remarkably outperforms previous approaches in terms of recovery accuracy.

We propose a saliency-maximized audio spectrogram as a representation that lets human analysts quickly search for and detect events in audio recordings. By rendering target events as visually salient patterns, this representation minimizes the time and effort needed to examine a recording. In particular, we propose a transformation of a conventional spectrogram that maximizes the mutual information between the spectrograms of isolated target events and the estimated saliency of the overall visual representation. When subjects are shown spectrograms that are saliency-maximized, they perform significantly better in a 1/10-real-time acoustic event detection task.

In this paper, we propose a kernel multi-metric learning algorithm for multi-channel transient acoustic signal classification. The proposed method learns a set of metrics jointly for multi-channel transient acoustic signals in a kernel-induced feature space to exploit the non-linearity of the data for improving the classification performance. An effective algorithm is developed for the task of learning multiple metrics in the kernel space. By learning the multiple metrics jointly within a single unified optimization framework, we can learn better metrics to integrate the multiple channels of the signal for a joint classification. Experimental results compared with classical as well as recent algorithms on real-world acoustic datasets verified the effectiveness of the proposed method.

The problem of community detection in social media has been widely studied in the social networking community in the context of the structure of the underlying graphs. Most community detection algorithms use the links between the nodes in order to determine the dense regions in the graph. These dense regions are the communities of social media in the graph. Such methods are typically based purely on the linkage structure of the underlying social media network. However, in many recent applications, edge content is available in order to provide better supervision to the community detection process. Many natural representations of edges in social interactions such as shared images and videos, user tags and comments are naturally associated with content on the edges. While some work has been done on utilizing node content for community detection, the presence of edge content presents unprecedented opportunities and flexibility for the community detection process. We will show that such edge content can be leveraged in order to greatly improve the effectiveness of the community detection process in social media networks. We present experimental results illustrating the effectiveness of our approach.

Lack of human prior knowledge is one of the main reasons that semantic gap still remains when it comes to automatic multimedia understanding. In this work, we exploit the ontological structure of target concepts and propose an universal ontological inference framework for image understanding. The framework explicitly utilizes subclass and co-occurrence relation to effectively refine the coarse concept detections. Moreover, we show how to automatically construct and learn the underlying ontology required by the framework. As can be shown by experiments, the result is an effective and robust algorithm that characterizes well the structure of the target concepts and outperforms the state-of-the-art methods.

Sparsity is an ubiquitous property exhibited by many natural real-world data such as images, which has been playing an important role in image and multi-media data processing. However, for many data, such as images, the sparsity pattern is not completely random, i.e., there are structures over the sparse coefficients. By exploiting this structure, we can model the data better and may further improve the performance of the recovery algorithm. In this paper, we exploit the structured sparsity of natural images for image deblurring application. Experimental results clearly demonstrate the effectiveness of the proposed approach.

In recent years, designing the coding and pooling structures in layered networks has been shown to be a useful method for learning high-level feature representations for visual data. Yet, such learning structures have not been extensively studied for audio signals. In this paper, we investigate the different pooling strategies based on the sparse coding scheme and propose a temporal pyramid pooling method to extract discriminative and shift-invariant feature representations. We demonstrate the superiority of our new feature representation over traditional features on the acoustic event classification task. Index Terms: sparse coding, pooling, acoustic event classification 1.

The problem of transfer learning has recently been of great interest in a variety of machine learning applications. In this paper, we examine a new angle to the transfer learning problem, where we examine the problem of distance function learning. Specifically, we focus on the problem of how our knowledge of distance functions in one domain can be transferred to a new domain. A good semantic understanding of the feature space is critical in providing the domain specific understanding for setting up good distance functions. Unfortunately, not all domains have feature representations which are equally interpretable. For example, in some domains such as text, the semantics of the feature representation are clear, as a result of which it is easy for a domain expert to set up distance functions for specific kinds of semantics. In the case of image data, the features are semantically harder to interpret, and it is harder to set up distance functions, especially for particular semantic criteria. In this paper, we focus on the problem of transfer learning as a way to close the semantic gap between different domains, and show how to use correspondence information between two domains in order to set up distance functions for the semantically more challenging domain.

The clustering of social media objects provides intrinsic understanding of the similarity relationships between documents, images, and their contextual sources. Both content and link structure provide important cues for an effective clustering algorithm of the underlying objects. While link information provides useful hints for improving the clustering process, it also contains a significant amount of noisy information. Therefore, a robust clustering algorithm is required to reduce the impact of noisy links. In order to address the aforementioned problems, we propose heterogeneous random fields to model the structure and content of social media networks. We design a probability measure on the social media networks which output a configuration of clusters that are consistent with both content and link structure. Furthermore, noisy links can also be detected, and their impact on the clustering algorithm can be significantly reduced. We conduct experiments on a real social media network and show the advantage of the method over other state-of-the-art algorithms.

The World-Wide Web provides tremendous resources in Multimedia Data, Ubiquitous Interconnection, and Storage/Computing Power. In this short note, we raise but not answer the question: Can the WWW help bridge the Semantic Gap in Computer Vision? Even if the answer turns out to be NO, we hope that by exploring these resources, we may gain a deeper understanding of the Semantic Gap challenge.

Nonnegative matrix factorization (NMF) is a popular technique for finding parts-based, linear representations of nonnegative data. It has been successfully applied in a wide range of applications such as pattern recognition, information retrieval, and computer vision. However, NMF is essentially an unsupervised method and cannot make use of label information. In this paper, we propose a novel semi-supervised matrix decomposition method, called Constrained Nonnegative Matrix Factorization (CNMF), which incorporates the label information as additional constraints. Specifically, we show how explicitly combining label information improves the discriminating power of the resulting matrix decomposition. We explore the proposed CNMF method with two cost function formulations and provide the corresponding update solutions for the optimization problems. Empirical experiments demonstrate the effectiveness of our novel algorithm in comparison to the state-of-the-art approaches through a set of evaluations based on real-world applications.

Social media networks contain both content and context-specific information. Most existing methods work with either of the two for the purpose of multimedia mining and retrieval. In reality, both content and context information are rich sources of information for mining, and the full power of mining and processing algorithms can be realized only with the use of a combination of the two. This paper proposes a new algorithm which mines both context and content links in social media networks to discover the underlying latent semantic space. This mapping of the multimedia objects into latent feature vectors enables the use of any off-the-shelf multimedia retrieval algorithms. Compared to the state-of-the-art latent methods in multimedia analysis, this algorithm effectively solves the problem of sparse context links by mining the geometric structure underlying the content links between multimedia objects. Specifically for multimedia annotation, we show that an effective algorithm can be developed to directly construct annotation models by simultaneously leveraging both context and content information based on latent structure between correlated semantic concepts. We conduct experiments on the Flickr data set, which contains user tags linked with images. We illustrate the advantages of our approach over the state-of-the-art multimedia retrieval techniques.

Content-based multimedia indexing, retrieval, and processing as well as multimedia databases demand the structuring of the media content (image, audio, video, text, etc.), one significant goal being to associate the identity of the content to the individual segments of the signals. In this paper, we specifically address the problem of speaker clustering, the task of assigning every speech utterance in an audio stream to its speaker. We offer a complete treatment to the idea of partially supervised speaker clustering, which refers to the use of our prior knowledge of speakers in general to assist the unsupervised speaker clustering process. By means of an independent training data set, we encode the prior knowledge at the various stages of the speaker clustering pipeline via 1) learning a speaker-discriminative acoustic feature transformation, 2) learning a universal speaker prior model, and 3) learning a discriminative speaker subspace, or equivalently, a speaker-discriminative distance metric. We study the directional scattering property of the Gaussian mixture model (GMM) mean supervector representation of utterances in the high-dimensional space, and advocate exploiting this property by using the cosine distance metric instead of the euclidean distance metric for speaker clustering in the GMM mean supervector space. We propose to perform discriminant analysis based on the cosine distance metric, which leads to a novel distance metric learning algorithm—linear spherical discriminant analysis (LSDA). We show that the proposed LSDA formulation can be systematically solved within the elegant graph embedding general dimensionality reduction framework. Our speaker clustering experiments on the GALE database clearly indicate that 1) our speaker clustering methods based on the GMM mean supervector representation and vector-based distance metrics outperform traditional speaker clustering methods based on the “bag of acoustic features” representation and statistical model-based distance metrics, 2) our advocated use of the cosine distance metric yields consistent increases in the speaker clustering performance as compared to the commonly used euclidean distance metric, 3) our partially supervised speaker clustering concept and strategies significantly improve the speaker clustering performance over the baselines, and 4) our proposed LSDA algorithm further leads to state-of-the-art speaker clustering performance.

The articles in this special issue focus on web-scale multimedia processing as well as applications for its use.

The abundance of multimedia data on the Web presents both challenges (how to annotate, search, and mine) and opportunities (crawling the Web to create large structured multimedia data bases which can be used to do inference effectively). Because of the huge data volume, considering all semantic concepts as on the same (flat) level is not viable. In this paper, we introduce a unified STRUCTURED representation called multimedia information networks (MINets), which incorporates ontology and cross-media links, covering both content and context knowledge. Ontology and cross-media structures are constructed and expanded by automatically constructing MINets from web-scale data by state-of-the-art information extraction and knowledge-based population techniques. The resultant MINet will contain a wide range of linkages, including logical, statistical, and semantic relations among informative concept nodes, which connects proliferative ontology as well as cross-media web-scale resources together. The raw data collected in construction phase often contain much noisy, incomplete, or even conflicting information which could be detrimental to information extraction and utilization. Then, the redundant link structure can be utilized to distill MINets and improve quality of information (QoI). Moreover, advanced inference theory and system can be built upon the linked MINets, and then high-level ontological knowledge can be inferred and integrated in a logically harmonious network structure in MINets which is consistent with human cognition. Even more, as information channels, the ontology and cross-media links in MINets connect informative knowledge resources together, which makes it possible to increase the portability of information between different resources to increase information utilization levels.

In this paper, we propose a novel coupled dictionary training method for single image super-resolution based on patchwise sparse recovery, where the learned couple dictionaries relate the low- and high-resolution image patch spaces via sparse representation. The learning process enforces that the sparse representation of a low-resolution image patch in terms of the low-resolution dictionary can well reconstruct its underlying high-resolution image patch with the dictionary in the highresolution image patch space. We model the learning problem as a bilevel optimization problem, where the optimization includes an 1-norm minimization problem in its constraints. Implicit differentiation is employed to calculate the desired gradient for stochastic gradient descent. We demonstrate that our coupled dictionary learning method can outperform the existing joint dictionary training method both quantitatively and qualitatively. Furthermore, for real applications, we speed up the algorithm approximately 10 times by learning a neural network model for fast sparse inference and selectively processing only those visually salient regions. Extensive experimental comparisons with stateof- the-art super-resolution algorithms validate the effectiveness of our proposed approach.

We propose a new single image super resolution (SR) algorithm via Bayesian modeling with a natural image prior modeled by a high-order Markov random field (MRF). SR is one of the long-standing and active topics in image processing community. It is of great use in many practical applications, such as astronomical observation, medical imaging, and the adaptation of low-resolution contents onto high-resolution displays. One category of the conventional approaches for image SR is formulating the problem with Bayesian modeling techniques and then obtaining its maximum-a-posteriori solution, which actually boils down to a regularized regression task. Although straightforward, this approach cannot exploit the full potential offered by the probabilistic modeling, as only the posterior mode is sought. On the other hand, current Bayesian SR approaches using the posterior mean estimation typically use very simple prior models for natural images to ensure the computational tractability. In this paper, we present a Bayesian image SR approach with a flexible high-order MRF model as the prior for natural images. The minimum mean square error (MMSE) criteria are used for estimating the HR image. A Markov chain Monte Carlo-based sampling algorithm is presented for obtaining the MMSE solution. The proposed method cannot only enjoy the benefits offered by the flexible prior, but also has the advantage of making use of the probabilistic modeling to perform a posterior mean estimation, thus is less sensitive to the local minima problem as the MAP solution. Experimental results indicate that the proposed method can generate competitive or better results than state-of-the-art SR algorithms.

With the popularity of digital cameras and camera phones, it is common for different people, who may or may not know each other, to attend the same event and take pictures and videos from different spatial or personal perspectives. Within the realm of social media, it is desirable to enable these people to select and share their pictures and videos in order to enrich memories and facilitate social networking. However, it is cumbersome to manually manage these photos from different cameras, of which the clocks settings are often not calibrated. In this paper, we propose automatic algorithms to address the above problems. First, we accurately align different photo streams or sequences from different photographers for the same event in chronological order on a common timeline, while respecting the time constraints within each photo stream. Given the preferred similarity measures (e.g., visual, and spatial similarities), our algorithm performs photo stream alignment via matching on a bipartite kernel sparse representation graph that forces the data connections to be sparse in an explicit fashion. Furthermore, we can produce a summary master stream from the aligned super stream of photos for efficient sharing by removing those redundant photos in the super stream while accounting for the temporal integrity. Based on a similar kernel sparse representation graph, our master stream summarization algorithm performs greedy backward selection to drop redundant photos without affecting the integrity of remaining photos for the entire event. We evaluate our algorithms on real-world personal online albums for 36 events and demonstrate its efficacy in automatically facilitating collaborative photo collection and sharing.

The four papers in this special issue focus on the novel design and methodology of subspace and manifold learning for image and video indexing and search.

This paper details the authors' efforts to push the baseline of emotion recognition performance on the Geneva Multimodal Emotion Portrayals (GEMEP) Facial Expression Recognition and Analysis database. Both subject-dependent and subject-independent emotion recognition scenarios are addressed in this paper. The approach toward solving this problem involves face detection, followed by key-point identification, then feature generation, and then, finally, classification. An ensemble of features consisting of hierarchical Gaussianization, scale-invariant feature transform, and some coarse motion features have been used. In the classification stage, we used support vector machines. The classification task has been divided into person-specific and person-independent emotion recognitions using face recognition with either manual labels or automatic algorithms. We achieve 100% performance for the person-specific one, 66% performance for the person-independent one, and 80% performance for overall results, in terms of classification rate, for emotion recognition with manual identification of subjects.

This paper investigates the joint-structured-sparsity-based methods for transient acoustic signal classification with multiple measurements. By joint structured sparsity, we not only use the sparsity prior for each measurement but we also exploit the structural information across the sparse representation vectors of multiple measurements. Several different sparse prior models are investigated in this paper to exploit the correlations among the multiple measurements with the notion of the joint structured sparsity for improving the classification accuracy. Specifically, we propose models with the joint structured sparsity under different assumptions: same sparse code model, common sparse pattern model, and a newly proposed joint dynamic sparse model. For the joint dynamic sparse model, we also develop an efficient greedy algorithm to solve it. Extensive experiments are carried out on real acoustic data sets, and the results are compared with the conventional discriminative classifiers in order to verify the effectiveness of the proposed method.

This paper explores the possibility of acquiring inverse li ght transport directly. The current strategy of obtaining an inverse light transport ma trix involves two steps: First, acquire the forward light transport matrix (f-LTM) and then calculate the inverse of the f-LTM. Both steps of the strategy requires considerable computational power. In addition to computational cost, the measurement error incurred at the first step inevitably propagates to or potentially gets amplified in the matrix inv ersion step. In this paper, we propose a sensing strategy that acquires the inverse light transport matrix (i-LTM) directly, without reconstructing the f-LTM. Our direct strategy reduces both computational error and cost of acquiring i-LTM. For that, we propose a compressive inverse theory. Following the compressible property of i-LTM, a reconstruction condition for i-LTM is introduced. This new framework implies a trade-off between two factors: condition numbers of submatrices of f-LTM and the isometry constant of the illumination pattern. Our direct i-LTM reconstruction method is then demonstrated with a 2nd-bounce separation experiment on an M-shaped panel scene. Finally by quantitatively comparing our method with the existing two-stage approach, our method shows higher accuracy with lower complexity. The proofs of main theorem/lemma are contained in the supplementary material. The compressive inverse theory is general and potentially useful for wider application.

Most research efforts on image classification so far have been focused on medium-scale datasets, which are often defined as datasets that can fit into the memory of a desktop (typically 4G~48G). There are two main reasons for the limited effort on large-scale image classification. First, until the emergence of ImageNet dataset, there was almost no publicly available large-scale benchmark data for image classification. This is mostly because class labels are expensive to obtain. Second, large-scale classification is hard because it poses more challenges than its medium-scale counterparts. A key challenge is how to achieve efficiency in both feature extraction and classifier training without compromising performance. This paper is to show how we address this challenge using ImageNet dataset as an example. For feature extraction, we develop a Hadoop scheme that performs feature extraction in parallel using hundreds of mappers. This allows us to extract fairly sophisticated features (with dimensions being hundreds of thousands) on 1.2 million images within one day. For SVM training, we develop a parallel averaging stochastic gradient descent (ASGD) algorithm for training one-against-all 1000-class SVM classifiers. The ASGD algorithm is capable of dealing with terabytes of training data and converges very fast-typically 5 epochs are sufficient. As a result, we achieve state-of-the-art performance on the ImageNet 1000-class classification, i.e., 52.9% in classification accuracy and 71.8% in top 5 hit rate.

In recent years, knowledge transfer algorithms have become one of most the active research areas in learning visual concepts. Most of the existing learning algorithms focuses on leveraging the knowledge transfer process which is specific to a given category. However, in many cases, such a process may not be very effective when a particular target category has very few samples. In such cases, it is interesting to examine, whether it is feasible to use cross-category knowledge for improving the learning process by exploring the knowledge in correlated categories. Such a task can be quite challenging due to variations in semantic similarities and differences between categories, which could either help or hinder the cross-category learning process. In order to address this challenge, we develop a cross-category label propagation algorithm, which can directly propagate the inter-category knowledge at instance level between the source and the target categories. Furthermore, this algorithm can automatically detect conditions under which the transfer process can be detrimental to the learning process. This provides us a way to know when the transfer of cross-category knowledge is both useful and desirable. We present experimental results on real image and video data sets in order to demonstrate the effectiveness of our approach.

Local classifiers have obtained great success in classification task due to its powerful discriminating ability on local regions. However, most of them still have restricted generalization in twofold: (1) each local classifier is sensitive to noise in local regions which leads to overfitting phenomenon in local classifiers; (2) the local classifiers also ignore the local correlation determined by the sample distribution in each local region. To overcome the above two problems, we present a novel locality-sensitive support vector machine (LSSVM) in this paper for image retrieval problem. This classifier applies locality-sensitive hashing (LSH) to divide the whole feature space into a number of local regions, on each of them a local model can be better constructed due to smaller within-class variation on it. To avoid these local models from overfitting into locality-sensitive structures, it imposes a global regularizer across local regions so that local classifiers are smoothly glued together to form a regularized overall classifier. local correlation is modeled to capture the sample distribution that determines the locality structure of each local region, which can increase the discriminating ability of the algorithm. To evaluate the performance, we apply the proposed algorithm into image retrieval task and competitive results are obtained on the real-world web image data set.

This paper proposes a new method for comparing 3D facial shapes using facial level curves. The pair- and segment-wise distances between the level curves comprise the spatio-temporal features for expression recognition from 3D dynamic faces. The paper further introduces universal background modeling and maximum a posteriori adaptation for hidden Markov models, leading to a decision boundary focus classification algorithm. Both techniques, when combined, yield a high overall recognition accuracy of 92.22% on the BU-4DFE database in our preliminary experiments. Noticeably, our feature extraction method is very efficient, requiring simple preprocessing, and robust to variations of the input data quality.

This work details the authors' efforts to push the baseline of expression recognition performance on a realistic database. Both subject-dependent and subject-independent emotion recognition scenarios are addressed in this work. These two happen frequently in real life settings. The approach towards solving this problem involves face detection, followed by key point identification, then feature generation and then finally classification. An ensemble of features comprising of Hierarchial Gaussianization (HG), Scale Invariant Feature Transform (SIFT) and Optic Flow have been incorporated. In the classification stage we used SVMs. The classification task has been divided into person specific and person independent emotion recognition. Both manual labels and automatic algorithms for person verification have been attempted. They both give similar performance.

In this paper, we present a novel joint sparse representation based method for acoustic signal classification with multiple measurements. The proposed method exploits the correlations among the multiple measurements with the notion of joint sparsity for improving the classification accuracy. Extensive experiments are carried out on real acoustic data sets and the results are compared with the conventional discriminative classifiers in order to verify the effectiveness of the proposed method.

We address the problem of visual recognition from multiple observations of the same physical object, which can be generated under different conditions, such as frames at different time instances or snapshots from different viewpoints. We formulate the multi-observation visual recognition task as a joint sparse representation model and take advantage of the correlations among the multiple observations for classification using a novel joint dynamic sparsity prior. The proposed joint dynamic sparsity prior promotes shared joint sparsity pattern among the multiple sparse representation vectors at class-level, while allowing distinct sparsity patterns at atom-level within each class in order to facilitate a flexible representation. The proposed method can handle both homogenous as well as heterogenous data within the same framework. Extensive experiments on various visual classification tasks including face recognition and generic object classification demonstrate that the proposed method outperforms existing state-of-the-art methods

Most previous visual recognition systems simply assume ideal inputs without real-world degradations, such as low resolution, motion blur and out-of-focus blur. In presence of such unknown degradations, the conventional approach first resorts to blind image restoration and then feeds the restored image into a classifier. Treating restoration and recognition separately, such a straightforward approach, however, suffers greatly from the defective output of the ill-posed blind image restoration. In this paper, we present a joint blind image restoration and recognition method based on the sparse representation prior to handle the challenging problem of face recognition from low-quality images, where the degradation model is realistic and totally unknown. The sparse representation prior states that the degraded input image, if correctly restored, will have a good sparse representation in terms of the training set, which indicates the identity of the test image. The proposed algorithm achieves simultaneous restoration and recognition by iteratively solving the blind image restoration in pursuit of the sparest representation for recognition. Based on such a sparse representation prior, we demonstrate that the image restoration task and the recognition task can benefit greatly from each other. Extensive experiments on face datasets under various degradations are carried out and the results of our joint model shows significant improvements over conventional methods of treating the two tasks independently.

This paper studies the problem of latent periodic topic analysis from time stamped documents. The examples of time stamped documents include news articles, sales records, financial reports, TV programs, and more recently, posts from social media websites such as Flickr, Twitter, and Face book. Different from detecting periodic patterns in traditional time series database, we discover the topics of coherent semantics and periodic characteristics where a topic is represented by a distribution of words. We propose a model called LPTA (Latent Periodic Topic Analysis) that exploits the periodicity of the terms as well as term co-occurrences. To show the effectiveness of our model, we collect several representative datasets including Seminar, DBLP and Flickr. The results show that our model can discover the latent periodic topics effectively and leverage the information from both text and time well.

A key problem in image annotation is to learn the underlying semantics. However, finding such semantic embeddings is a challenge task and often requires large amount of tagging information. In this paper, we propose to utilize multi-modality cues by incorporating visual and textual information as embedded objects. The paper further presents a multi-task learning framework that simultaneously learns the approximation of two semantic embeddings with efficient multi-stage convex relaxation technique. The experiments show that the proposed method presents very promising performance in both memory usage and training time for large-scale dataset, as well as image classification accuracy.

We consider the problem of automatically recognizing a human face from its multi-view images with unconstrained poses and illuminations. We formulate the multi-view face recognition problem as that of classifying among several multi-input (views) regression models by using a novel joint dynamic sparse representation method which exploits jointly the inter-correlation among all the multi-view images in order to make a decision. Extensive experiments on CMU Multi-PIE face database are conducted to verify the efficacy of the proposed method.

In this paper, we propose an extension of the Non-Local Kernel Regression (NL-KR) method and apply it to super-resolution (SR) tasks. The proposed method extends NL-KR via generalizing the self-similarity from single-scale to multi-scale, and propose an effective SR algorithm using the proposed multi-scale NL-KR model. Experimental results on both synthetic and real images demonstrate the effectiveness of the proposed method.

We explore event recognition in the personal photo album setting where the task is not to identify event in individual photo but in the whole album. This setting arises from the way people organize their own collections and the fact that individual photo in these albums often fails to convey meaningful event semantic behind the the album. We work on this problem in a object-centric manner, i.e. we train detectors for objects relevant to the events in the holiday dataset we built, and then detect these holidays based on object detector outputs. The prior knowledge, i.e. what objects are relevant to the event, is obtained statistically from mass image collection web site and thus tends to be more accurate and less biased.

In this work, we propose a novel supervised matrix factorization method used directly as a multi-class classifier. The coefficient matrix of the factorization is enforced to be sparse by ℓ1-norm regularization. The basis matrix is composed of atom dictionaries from different classes, which are trained in a jointly supervised manner by penalizing inhomogeneous representations given the labeled data samples. The learned basis matrix models the data of interest as a union of discriminative linear subspaces by sparse projection. The proposed model is based on the observation that many high-dimensional natural signals lie in a much lower dimensional subspaces or union of subspaces. Experiments conducted on several datasets show the effectiveness of such a representation model for classification, which also suggests that a tight reconstructive representation model could be very useful for discriminant analysis.

We propose a sparse representation based blind image deblurring method. The proposed method exploits the sparsity property of natural images, by assuming that the patches from the natural images can be sparsely represented by an over-complete dictionary. By incorporating this prior into the deblurring process, we can effectively regularize the ill-posed inverse problem and alleviate the undesirable ring effect which is usually suffered by conventional deblurring methods. Experimental results compared with state-of-the-art blind deblurring method demonstrate the effectiveness of the proposed method.

Super resolution is of great use in many visual media related scenarios, such as displaying low resolution contents on High-Definition TV(HD-TV). In these scenarios, the efficiency of the super resolution process is of vital importance. This paper presents a fast learning based super-resolution method. The proposed method speeds up the sparse representation based super-resolution method by learning a dual dictionary, and replaces the sparse recovery step by simple matrix multiplication, which is much more computationally efficient. Experiments demonstrate that the proposed method can generate desirable super-resolved images with significant computational advantages.

In this paper, we study the problem of recognizing events in personal photo albums. In consumer photo collections or online photo communities, photos are usually organized in albums according to their events. However, interpreting photo albums is more complicated than the traditional problem of understanding single photos, because albums generally exhibit much more varieties than single image. To solve this challenge, we propose a novel representation, called Compositional Object Pattern, which characterizes object level pattern conveying much richer semantic than low level visual feature. To interpret the rich semantics in albums, we mine frequent object patterns in the training set, and then rank them by their discriminating power. The album feature is then set as the frequencies of these frequent and discriminative patterns, called Compositional Object Pattern Frequency(COPF). We show with experimental result that our algorithm is capable of recognizing holidays with accuracy higher than the baseline method.

High dimensional similarity search in large scale databases becomes an important challenge due to the advent of Internet. For such applications, specialized data structures are required to achieve computational efficiency. Traditional approaches relied on algorithmic constructions that are often data independent (such as Locality Sensitive Hashing) or weakly dependent (such as kd-trees, k-means trees). While supervised learning algorithms have been applied to related problems, those proposed in the literature mainly focused on learning hash codes optimized for compact embedding of the data rather than search efficiency. Consequently such an embedding has to be used with linear scan or another search algorithm. Hence learning to hash does not directly address the search efficiency issue. This paper considers a new framework that applies supervised learning to directly optimize a data structure that supports efficient large scale search. Our approach takes both search quality and computational cost into consideration. Specifically, we learn a boosted search forest that is optimized using pair-wise similarity labeled examples. The output of this search forest can be efficiently converted into an inverted indexing data structure, which can leverage modern text search infrastructure to achieve both scalability and efficiency. Experimental results show that our approach significantly outperforms the start-of-the-art learning to hash methods (such as spectral hashing), as well as state-of-the-art high dimensional search algorithms (such as LSH and k-means trees).

We propose PhotoNet, a picture delivery service for camera sensor networks. PhotoNet is motivated by the needs of disaster-response applications, where a group of survivors and first responders may survey damage and send images to a rescue center in the absence of a functional communication infrastructure. The protocol runs on mobile devices, handling opportunistic forwarding (when they come in contact) and in-network storage. It assigns priorities to images for forwarding and replacement depending on the degree of similarity (or dissimilarity) among them, such that scarce resources are assigned to delivery of most ``deserving'' content first. Prioritization aims at reducing semantic redundancy such as that between pictures of the same scene at the same location taken from slightly different angles. This is in contrast to redundancy among identical objects and among time series data. PhotoNet delivers more diverse pictures in terms of event coverage suppressing logically redundant content belonging to the same event. We show that, in resource constrained networks, reducing semantic redundancy can significantly improve the utility of the service.

Social media such as those residing in the popular photo sharing websites is attracting increasing attention in recent years. As a type of user-generated data, wisdom of the crowd is embedded inside such social media. In particular, millions of users upload to Flickr their photos, many associated with temporal and geographical information. In this paper, we investigate how to rank the trajectory patterns mined from the uploaded photos with geotags and timestamps. The main objective is to reveal the collective wisdom recorded in the seemingly isolated photos and the individual travel sequences reflected by the geo-tagged photos. Instead of focusing on mining frequent trajectory patterns from geo-tagged social media, we put more effort into ranking the mined trajectory patterns and diversifying the ranking results. Through leveraging the relationships among users, locations and trajectories, we rank the trajectory patterns. We then use an exemplar-based algorithm to diversify the results in order to discover the representative trajectory patterns. We have evaluated the proposed framework on 12 different cities using a Flickr dataset and demonstrated its effectiveness.

In this paper, we study the problem of transfer learning from text to images in the context of network data in which link based bridges are available to transfer the knowledge between the different domains. The problem of classification of image data is often much more challenging than text data because of the following two reasons: (a) Labeled text data is very widely available for classification purposes. On the other hand, this is often not the case for image data, in which a lot of images are available from many sources, but many of them are often not labeled. (b) The image features are not directly related to semantic concepts inherent in class labels. On the other hand, since text data tends to have natural semantic interpretability (because of their human origins), they are often more directly related to class labels. Therefore, the relationships between the images and text features also provide additional hints for the classification process in terms of the image feature transformations which provide the most effective results.

This paper studies the problem of discovering and comparing geographical topics from GPS-associated documents. GPS-associated documents become popular with the pervasiveness of location-acquisition technologies. For example, in Flickr, the geo-tagged photos are associated with tags and GPS locations. In Twitter, the locations of the tweets can be identified by the GPS locations from smart phones. Many interesting concepts, including cultures, scenes, and product sales, correspond to specialized geographical distributions. In this paper, we are interested in two questions: (1) how to discover different topics of interests that are coherent in geographical regions? (2) how to compare several topics across different geographical locations? To answer these questions, this paper proposes and compares three ways of modeling geographical topics: location-driven model, text-driven model, and a novel joint model called LGTA (Latent Geographical Topic Analysis) that combines location and text. To make a fair comparison, we collect several representative datasets from Flickr website including Landscape, Activity, Manhattan, National park, Festival, Car, and Food. The results show that the first two methods work in some datasets but fail in others. LGTA works well in all these datasets at not only finding regions of interests but also providing effective comparisons of the topics across different locations. The results confirm our hypothesis that the geographical distributions can help modeling topics, while topics provide important cues to group different geographical regions.

Matrix factorization techniques have been frequently applied in information retrieval, computer vision, and pattern recognition. Among them, Nonnegative Matrix Factorization (NMF) has received considerable attention due to its psychological and physiological interpretation of naturally occurring data whose representation may be parts based in the human brain. On the other hand, from the geometric perspective, the data is usually sampled from a low-dimensional manifold embedded in a high-dimensional ambient space. One then hopes to find a compact representation,which uncovers the hidden semantics and simultaneously respects the intrinsic geometric structure. In this paper, we propose a novel algorithm, called Graph Regularized Nonnegative Matrix Factorization (GNMF), for this purpose. In GNMF, an affinity graph is constructed to encode the geometrical information and we seek a matrix factorization, which respects the graph structure. Our empirical study shows encouraging results of the proposed algorithm in comparison to the state-of-the-art algorithms on real-world problems.

The motivations for organizing this special section were to better address the challenges of face recognition in real-world scenarios, to promote systematic research and evaluation of promising methods and systems, to provide a snapshot of where we are in this domain, and to stimulate discussion about future directions. We solicited original contributions of research on all aspects of real-world face recognition, including: the design of robust face similarity features and metrics; robust face clustering and sorting algorithms; novel user interaction models and face recognition algorithms for face tagging; novel applications of web face recognition; novel computational paradigms for face recognition; challenges in large scale face recognition tasks, e.g., on the Internet; face recognition with contextual information; face recognition benchmarks and evaluation methodology for moderately controlled or uncontrolled environments; and video face recognition. We received 42 original submissions, four of which were rejected without review; the other 38 papers entered the normal review process. Each paper was reviewed by three reviewers who are experts in their respective topics. More than 100 expert reviewers have been involved in the review process. The papers were equally distributed among the guest editors. A final decision for each paper was made by at least two guest editors assigned to it. To avoid conflict of interest, no guest editor submitted any papers to this special section.

We consider the active learning problem, which aims to select the most representative points. Out of many existing active learning techniques, optimum experimental design (OED) has received considerable attention recently. The typical OED criteria minimize the variance of the parameter estimates or predicted value. However, these methods see only global euclidean structure, while the local manifold structure is ignored. For example, I-optimal design selects those data points such that other data points can be best approximated by linear combinations of all the selected points. In this paper, we propose a novel active learning algorithm which takes into account the local structure of the data space. That is, each data point should be approximated by the linear combination of only its neighbors. Given the local reconstruction coefficients for every data point and the coordinates of the selected points, a transductive learning algorithm called Locally Linear Reconstruction (LLR) is proposed to reconstruct every other point. The most representative points are thus defined as those whose coordinates can be used to best reconstruct the whole data set. The sequential and convex optimization schemes are also introduced to solve the optimization problem. The experimental results have demonstrated the effectiveness of our proposed method.

Detecting text and caption from videos is important and in great demand for video retrieval, annotation, indexing, and content analysis. In this paper, we present a corner based approach to detect text and caption from videos. This approach is inspired by the observation that there exist dense and orderly presences of corner points in characters, especially in text and caption. We use several discriminative features to describe the text regions formed by the corner points. The usage of these features is in a flexible manner, thus, can be adapted to different applications. Language independence is an important advantage of the proposed method. Moreover, based upon the text features, we further develop a novel algorithm to detect moving captions in videos. In the algorithm, the motion features, extracted by optical flow, are combined with text features to detect the moving caption patterns. The decision tree is adopted to learn the classification criteria. Experiments conducted on a large volume of real video shots demonstrate the efficiency and robustness of our proposed approaches and the real-world system. Our text and caption detection system was recently highlighted in a worldwide multimedia retrieval competition, Star Challenge, by achieving the superior performance with the top ranking.

In this article, beyond solo-activity analysis for single object, we study the more complicated pair-activity recognition problem by exploring the relationship between two active objects based on their trajectory clues obtained from video sensor. Our contributions are three-fold. First, we design two sets of features for representing the pair-activities encoded as length-variable trajectory pairs. One set characterizes the strength of causality between two trajectories, for example, the causality ratio and feedback ratio based on the Granger Causality Test (GCT), and another set describes the style of causality between two trajectories, for example, the sampled frequency responses of the digital filter with these two trajectories as the input and output discrete signals respectively. These features along with conventional velocity and position features of a trajectory-pair are essentially of multi-modalities, and may be greatly different in scales and importance. To make full use of them, we then develop a novel feature fusing procedure to learn the coefficients for weighting these features by maximizing the discriminating power measured by weighted correlation. Finally, we collected a pair-activity database of five popular categories, each of which consists of about 170 instances. The extensive experiments on this database validate the effectiveness of the designed features for pair-activity representation, and also demonstrate that the proposed feature fusing procedure significantly boosts the pair-activity classification accuracy.

In recent years, many research works have been carried out to recognize human actions from video clips. To learn an effective action classifier, most of the previous approaches rely on enough training labels. When being required to recognize the action in a different dataset, these approaches have to re-train the model using new labels. However, labeling video sequences is a very tedious and time-consuming task, especially when detailed spatial locations and time durations are required. In this paper, we propose an adaptive action detection approach which reduces the requirement of training labels and is able to handle the task of cross-dataset action detection with few or no extra training labels. Our approach combines model adaptation and action detection into a Maximum a Posterior (MAP) estimation framework, which explores the spatial-temporal coherence of actions and makes good use of the prior information which can be obtained without supervision. Our approach obtains state-of-the-art results on KTH action dataset using only 50% of the training labels in tradition approaches. Furthermore, we show that our approach is effective for the cross-dataset detection which adapts the model trained on KTH to two other challenging datasets.

Due to the lack of explicit spatial consideration, existing epitome model may fail for image recognition and target detection, which directly motivates us to propose the so-called spatialized epitome in this paper. Extended from the original graphical model of epitome, the spatialized epitome provides a general framework to integrate both appearance and spatial arrangement of patches in the image to achieve a more precise likelihood representation for image(s) and eliminate ambiguities in image reconstruction and recognition. From the extended graphical model of epitome, an EM learning procedure is derived under the framework of variational approximation. The learning procedure can generate an optimized summary of the image appearance with spatial distribution of the similar patches. From the spatialized epitome, we present a principled way of inferring the probability of a new input image under the learnt model and thereby enabling image recognition and target detection. We show how the incorporation of spatial information enhances the epitome's ability for discrimination on several vision tasks, e.g., misalignment/cross-pose face recognition and vehicle detection with a few training samples.

The traditional SPM approach based on bag-of-features (BoF) requires nonlinear classifiers to achieve good image classification performance. This paper presents a simple but effective coding scheme called Locality-constrained Linear Coding (LLC) in place of the VQ coding in traditional SPM. LLC utilizes the locality constraints to project each descriptor into its local-coordinate system, and the projected coordinates are integrated by max pooling to generate the final representation. With linear classifier, the proposed approach performs remarkably better than the traditional nonlinear SPM, achieving state-of-the-art performance on several benchmarks. Compared with the sparse coding strategy [22], the objective function used by LLC has an analytical solution. In addition, the paper proposes a fast approximated LLC method by first performing a K-nearest-neighbor search and then solving a constrained least square fitting problem, bearing computational complexity of O(M + K2). Hence even with very large codebooks, our system can still process multiple frames per second. This efficiency significantly adds to the practical values of LLC for real applications.

In this paper, we propose a novel supervised hierarchical sparse coding model based on local image descriptors for classification tasks. The supervised dictionary training is performed via back-projection, by minimizing the training error of classifying the image level features, which are extracted by max pooling over the sparse codes within a spatial pyramid. Such a max pooling procedure across multiple spatial scales offer the model translation invariant properties, similar to the Convolutional Neural Network (CNN). Experiments show that our supervised dictionary improves the performance of the proposed model significantly over the unsupervised dictionary, leading to state-of-the-art performance on diverse image databases. Further more, our supervised model targets learning linear features, implying its great potential in handling large scale datasets in real applications.

This work aims to build a system to suggest tourist destinations based on visual matching and minimal user input. A user can provide either a photo of the desired scenary or a keyword describing the place of interest, and the system will look into its database for places that share the visual characteristics. To that end, we first cluster a large-scale geotagged web photo collection into groups by location and then find the representative images for each group. Tourist destination recommendations are produced by comparing the query against the representative tags or representative images under the premise of "if you like that place, you may also like these places".

One important class of state emission densities of the hiddenMarkov model (HMM) is the Gaussian mixture densities. The classical Baum-Welch algorithm often fails to reliably learn the Gaussian mixture densities when there is insufficient training data, due to the large number of free parameters present in the model. In this paper, we propose a novel strategy for robustly and accurately learning the Gaussian mixture state emission densities of the HMM. The strategy is based on an ensemble framework for probability density estimation in which the learning of the Gaussian mixture densities is formulated as a gradient descent search in a function space. The resulting learning algorithm is called “the boosting Baum-Welch algorithm.” Our preliminary experiment results on emotion recognition from speech show that the proposed algorithm outperforms the original Baum-Welch algorithm on this task.

In this paper, we propose a novel algorithm for reconstructing the 3D shape and texture of human faces from two stereo images, which are captured from calibrated cameras. Our approach works in a sparse to dense manner: we first build a coarse shape estimation based on 3D keypoints, and then use a linear morphable model to efficiently match the detail shape and texture. Compared with the previous works, our algorithm can reconstruct the 3D face shape in a speed comparable with that of the fastest algorithm available, but gives a higher accuracy. It can also recover the texture with more complete, realistic looking. Our results show that the new algorithm possesses significant characteristics of a 3D face model reconstruction system, and is especially useful for face recognition and animation applications in practice.

This paper presents a novel hierarchical density estimation approach for image classification. We first build a collection of randomized decision trees in a discriminative way to split the feature space into small regions. Then for each region, class-conditional Gaussians are learnt to characterize the “local” distribution of feature vectors falling into that region. The parameters of the Gaussians are reliably estimated through hierarchical maximum a posteriori (MAP) estimation and smoothed across multiple randomized trees in the forest. Compared with the widely-used Gaussian Mixture Models (GMM), our new approach not only yields more reliable parameter estimation, but also greatly reduces the computational cost at the testing stage. Experiments on scene classification demonstrate the effectiveness and efficiency of our new approach.

Inspired from the observation that license plates are very salient to human visual perception, we propose a novel license plate detection algorithm based on image saliency in this paper. The proposed algorithm consists of two parts. The first part segments out the characters on a license plate using an intensity saliency map with a high recall rate. The second part applies a sliding window on these characters to compute some saliency-related features to detect license plates. We test the robustness of our algorithm by applying it on a mixed data set with high diversity collected from four databases. The mixed data set has 1024 images composed of license plates of all states of the U.S. We achieve a detection rate of 90% with False Positive Per Image (FPPI) = 12%. The detection box given by our algorithm has high precision, which will be very helpful for many applications such as license plate recognition.

In this paper, we propose a novel image representation for scene classification. Firstly, we model multiple order statistics of image patches via Gaussian Mixture Model(GMM) in a Bayesian framework. Secondly, we combine the information of mean and covariance of the GMM and represent it as a mean-covariance supervector through a new distance metric. Experimental results demonstrate that our new representation, by just using nearest centroid classifier, has significantly outperformed all existing methods on the fifteen scene category database.

This paper considers the problem of detecting actions from cluttered videos. Compared with the classical action recognition problem, this paper aims to estimate not only the scene category of a given video sequence, but also the spatial-temporal locations of the action instances. In recent years, many feature extraction schemes have been designed to describe various aspects of actions. However, due to the difficulty of action detection, e.g., the cluttered background and potential occlusions, a single type of features cannot solve the action detection problems perfectly in cluttered videos. In this paper, we attack the detection problem by combining multiple Spatial-Temporal Interest Point (STIP) features, which detect salient patches in the video domain, and describe these patches by feature of local regions. The difficulty of combining multiple STIP features for action detection is two folds: First, the number of salient patches detected by different STIP methods varies across different salient patches. How to combine such features is not considered by existing fusion methods. Second, the detection in the videos should be efficient, which excludes many slow machine learning algorithms. To handle these two difficulties, we propose a new approach which combines Gaussian Mixture Model with Branch-and-Bound search to efficiently locate the action of interest. We build a new challenging dataset for our action detection task, and our algorithm obtains impressive results. On classical KTH dataset, our method outperforms the state-of-the-art methods.

Automatic facial expression recognition from non-frontal views is a challenging research topic which has recently started to attract the attention of the research community. In this paper, we propose a novel approach to tackling this problem based on the ergodic hidden Markov model (EHMM) supervector representation of facial images. First, the scale-invariant feature transform (SIFT) feature vectors are extracted from a dense grid of every facial images. Next, an EHMM is trained over all facial images in the training set and is referred to as the universal background model (UBM). The UBM is then maximum a posteriori adapted to each facial image in the training and test sets to produce the image-specific EHMMs. Based on these EHMMs, we derive a supervector representation of the facial images by means of an upper bound approximation of the Kullback-Leibler divergence rate between two EHMMs. Finally, facial expression recognition is performed in the linear discriminant subspace of the EHMM supervectors using the k-nearest-neighbor classification algorithm. Our experiments of recognizing six universal facial expressions over extensive multiview facial images with seven pan angles (-45° ~ +45°) and five tilt angles (-30° ~ +30°), which are synthesized from the BU-3DFE facial expression database, show promising results compared to the state of the arts recently reported.

Classifying images using features extracted from densely sampled local patches has enjoyed significant success in many detection and recognition tasks. It is also well known that generally more than one type of feature is needed to achieve robust classification performance. Previous works using multiple features have addressed this issue either through simple concatenation of feature vectors or through combining feature specific kernels at the classifier level. In this work we introduce a novel approach for combining features at the feature level by projecting two types of features onto two respective subspaces in which they are maximally correlated. We use their correlation as an augmented feature and demonstrate improvement in classification accuracy over simple combination through concatenation in a pedestrian detection framework.

In the field of machine learning and pattern recognition, an alternative to conventional classification is hierarchical classification that exploits hierarchical relations between concepts of interest. To the best of our knowledge, all hierarchical classification methods in the literature are designed to reduce computation complexity without sacrificing too much on accuracy performance. In this work on image classification, we first propose a hierarchical image feature extraction that extracts image feature based on the location of current node in hierarchy to fit the images under current node and to better distinguish its subclasses. As far as we know, such node-dependent feature extraction has not been considered in the literature. Contrary to former hierarchical classification methods that only consider local structure of the hierarchy, we propose a novel cross-level hierarchical classification method that utilizes both global and local concept structures throughout the entire path decision-making process. Our experimental result on two datasets shows that the proposed hierarchical feature extraction combined with our novel hierarchical classification achieves better accuracy performance than conventional non-hierarchical classification methods, and hence conventional hierarchical methods as well.

With the explosive growth of digital cameras and online media, it has become crucial to design efficient methods that help users browse and search large image collections. The recent VisualRank algorithm [4] employs visual similarity to represent the link structure in a graph so that the classic PageRank algorithm can be applied to select the most relevant images. However, measuring visual similarity is difficult when there exist diversified semantics in the image collection, and the results from VisualRank cannot supply good visual summarization with diversity. This paper proposes to rank the images in a structural fashion, which aims to discover the diverse structure embedded in photo collections, and rank the images according to their similarity among local neighborhoods instead of across the entire photo collection. We design a novel algorithm named RankCompete, which generalizes the PageRank algorithm for the task of simultaneous ranking and clustering. The experimental results show that RankCompete outperforms VisualRank and provides an efficient but effective tool for organizing web photos.

Human age, as an important personal trait, can be directly inferred by distinct patterns emerging from the facial appearance. Derived from rapid advances in computer graphics and machine vision, computer-based age synthesis and estimation via faces have become particularly prevalent topics recently because of their explosively emerging real-world applications, such as forensic art, electronic customer relationship management, security control and surveillance monitoring, biometrics, entertainment, and cosmetology. Age synthesis is defined to rerender a face image aesthetically with natural aging and rejuvenating effects on the individual face. Age estimation is defined to label a face image automatically with the exact age (year) or the age group (year range) of the individual face. Because of their particularity and complexity, both problems are attractive yet challenging to computer-based application system designers. Large efforts from both academia and industry have been devoted in the last a few decades. In this paper, we survey the complete state-of-the-art techniques in the face image-based age synthesis and estimation topics. Existing models, popular algorithms, system performances, technical difficulties, popular face aging databases, evaluation protocols, and promising future directions are also provided with systematic discussions.

Image is a powerful medium for expressing one's ideas and rightly confirms the adage, “One picture is worth a thousand words.” In this work, we explore the application of world knowledge in the form of Wikipedia to achieve this objective-literally. In the first part, we disambiguate and rank semantic concepts associated with ambiguous keywords by exploiting link structure of articles in Wikipedia. In the second part, we explore an image representation in terms of keywords which reflect the semantic content of an image. Our approach is inspired by the desire to augment low-level image representation with massive amounts of world knowledge, to facilitate computer vision tasks like image retrieval based on this information. We represent an image as a weighted mixture of a predetermined set of concrete concepts whose definition has been agreed upon by a wide variety of audience. To achieve this objective, we use concepts defined by Wikipedia articles, e.g., sky, building, or automobile. An important advantage of our approach is availability of vast amounts of highly organized human knowledge in Wikipedia. Wikipedia evolves rapidly steadily increasing its breadth and depth over time.

Techniques from sparse signal representation are beginning to see significant impact in computer vision, often on nontraditional applications where the goal is not just to obtain a compact high-fidelity representation of the observed signal, but also to extract semantic information. The choice of dictionary plays a key role in bridging this gap: unconventional dictionaries consisting of, or learned from, the training samples themselves provide the key to obtaining state-of-the-art results and to attaching semantic meaning to sparse signal representations. Understanding the good performance of such unconventional dictionaries in turn demands new algorithmic and analytical techniques. This review paper highlights a few representative examples of how the interaction between sparse signal representation and computer vision can enrich both fields, and raises a number of open questions for further study.

In this letter, we propose a novel vector representation of stochastic signals for pattern recognition (PR) based on adapted ergodic hidden Markov models (HMMs). This vector representation is generic in nature and may be used with various types of stochastic signals (e.g., image, speech, etc.) and applied to a broad range of PR tasks (e.g., classification, regression, etc.). More importantly, by combining the vector representation with optimal distance metric learning (e.g., linear discriminant analysis) directly from the data, the performance of a PR system may be significantly improved. Our experiments on an image-based recognition task clearly demonstrate the effectiveness of the proposed vector representation of stochastic signals for potential use in many PR systems.

We consider the problem of estimating 3-D human body pose from visual signals within a discriminative framework. It is challenging because there is a wide gap between complex 3-D human motion and planar visual observation, which makes this a severely ill-conditioned problem. In this paper, we focus on three critical factors to tackle human body pose estimation, namely, feature extraction, learning algorithm, and camera utilization. On the feature level, we describe images using the salient interest points represented by scale-invariant feature transform (SIFT)-like descriptors, in which the position, appearance, and local structural information are encoded simultaneously. On the learning algorithm level, we propose to use Gaussian processes and multiple linear (ML) regression to model the mapping between poses and features. Fusing image information from multiple cameras in different views is of great interest to us on the camera level. We make a comprehensive evaluation on the HumanEva database and get two meaningful insights into the three crucial aspects for human pose estimation: 1) although the choice of feature is very important to the problem, once the learning algorithm becomes efficient, the choice of feature is no longer critical, and 2) the impact of information combination from multiple cameras on pose estimation is closely related to not only the quantity of image information, but also its quality. In most cases, it is true that the more information is involved, the better results can be achieved. But when the information quantity is the same, the differences in quality will lead to totally different performance. Furthermore, dense evaluations demonstrate that our approach is an accurate and robust solution to the human body pose estimation problem.

The graph construction procedure essentially determines the potentials of those graph-oriented learning algorithms for image analysis. In this paper, we propose a process to build the so-called directed l1-graph, in which the vertices involve all the samples and the ingoing edge weights to each vertex describe its l1-norm driven reconstruction from the remaining samples and the noise. Then, a series of new algorithms for various machine learning tasks, e.g., data clustering, subspace learning, and semi-supervised learning, are derived upon the l1-graphs. Compared with the conventional k-nearest-neighbor graph and epsilon-ball graph, the l1-graph possesses the advantages: (1) greater robustness to data noise, (2) automatic sparsity, and (3) adaptive neighborhood for individual datum. Extensive experiments on three real-world datasets show the consistent superiority of l1-graph over those classic graphs in data clustering, subspace learning, and semi-supervised learning tasks.

Subspace learning techniques for face recognition have been widely studied in the past three decades. In this paper, we study the problem of general subspace-based face recognition under the scenarios with spatial misalignments and/or image occlusions. For a given subspace derived from training data in a supervised, unsupervised, or semi-supervised manner, the embedding of a new datum and its underlying spatial misalignment parameters are simultaneously inferred by solving a constrained l1 norm optimization problem, which minimizes the l1 error between the misalignment-amended image and the image reconstructed from the given subspace along with its principal complementary subspace. A byproduct of this formulation is the capability to detect the underlying image occlusions. Extensive experiments on spatial misalignment estimation, image occlusion detection, and face recognition with spatial misalignments and/or image occlusions all validate the effectiveness of our proposed general formulation for misalignment-robust face recognition.

This paper presents a new approach to single-image super-resolution, based on sparse signal representation. Research on image statistics suggests that image patches can be well-represented as a sparse linear combination of elements from an appropriately chosen over-complete dictionary. Inspired by this observation, we seek a sparse representation for each patch of the low-resolution input, and then use the coefficients of this representation to generate the high-resolution output. Theoretical results from compressed sensing suggest that under mild conditions, the sparse representation can be correctly recovered from the downsampled signals. By jointly training two dictionaries for the low- and high-resolution image patches, we can enforce the similarity of sparse representations between the low resolution and high resolution image patch pair with respect to their own dictionaries. Therefore, the sparse representation of a low resolution image patch can be applied with the high resolution image patch dictionary to generate a high resolution image patch. The learned dictionary pair is a more compact representation of the patch pairs, compared to previous approaches, which simply sample a large amount of image patch pairs, reducing the computational cost substantially. The effectiveness of such a sparsity prior is demonstrated for both general image super-resolution and the special case of face hallucination. In both cases, our algorithm generates high-resolution images that are competitive or even superior in quality to images produced by other similar SR methods. In addition, the local sparse modeling of our approach is naturally robust to noise, and therefore the proposed algorithm can handle super-resolution with noisy inputs in a more unified framework.

This paper proposes a general framework for detecting unsafe states of a system whose basic real-time parameters are captured by multiple sensors. Our approach is to learn a danger-level function that can be used to alert the users of dangerous situations in advance so that certain measures can be taken to avoid the collapse. The main challenge to this learning problem is the labeling issue, i.e., it is difficult to assign an objective danger level at each time step to the training data, except at the collapse points, where a definitive penalty can be assigned, and at the successful ends, where a certain reward can be assigned. In this paper, we treat the danger level as an expected future reward (a penalty is regarded as a negative reward) and use temporal difference (TD) learning to learn a function for approximating the expected future reward, given the current and historical sensor readings. The TD learning obtains the approximation by propagating the penalties/rewards observable at collapse points or successful ends to the entire feature space following some constraints. This avoids the labeling issue and naturally allows a general framework to detect unsafe states. Our approach is applied to, but not limited to, the application of monitoring driving safety, and the experimental results demonstrate the effectiveness of the approach.

The 11 papers in this special issue can be categorized into five groups: Emotional speech synthesis and recognition; affective video content analysis; facial expressions and head movements; affect analysis in small groups; and audio-visual affective corpus.

While a sound spoken is described by a handful of frame-level spectral vectors, not all frames have equal contribution for either human perception or machine classification. In this paper, we introduce a novel framework to automatically emphasize important speech frames relevant to phonetic information. We jointly learn the importance of speech frames by a distance metric across the phone classes, attempting to satisfy a large margin constraint: the distance from a segment to its correct label class should be less than the distance to any other phone class by the largest possible margin. Furthermore, an universal background model structure is proposed to give the correspondence between statistical models of phone types and tokens, allowing us to use statistical models of each phone token in a large margin speech recognition framework. Experiments on TIMIT database demonstrated the effectiveness of our framework.

We investigate the biologically inspired features (BIF) for human age estimation from faces. As in previous bio-inspired models, a pyramid of Gabor filters are used at all positions of the input image for the S1units. But unlike previous models, we find that the pre-learned prototypes for the S2layer and then progressing to C2cannot work well for age estimation. We also propose to use Gabor filters with smaller sizes and suggest to determine the number of bands and orientations in a problem-specific manner, rather than using a predefined number. More importantly, we propose a new operator “STD” to encode the aging subtlety on faces. Evaluated on the large database YGA with 8,000 face images and the public available FG-NET database, our approach achieves significant improvements in age estimation accuracy over the state-of-the-artmethods. By applying our system to some Internet face images, we show the robustness of our method and the potential of cross-race age estimation, which has not been explored by any studies before.

Recently SVMs using spatial pyramid matching (SPM) kernel have been highly successful in image classification. Despite its popularity, these nonlinear SVMs have a complexity O(n2∼ n3) in training and O(n) in testing, where n is the training size, implying that it is nontrivial to scaleup the algorithms to handlemore than thousands of training images. In this paper we develop an extension of the SPM method, by generalizing vector quantization to sparse coding followed by multi-scale spatial max pooling, and propose a linear SPM kernel based on SIFT sparse codes. This new approach remarkably reduces the complexity of SVMs to O(n) in training and a constant in testing. In a number of image categorization experiments, we find that, in terms of classification accuracy, the suggested linear SPM based on sparse coding of SIFT descriptors always significantly outperforms the linear SPM kernel on histograms, and is even better than the nonlinear SPM kernels, leading to state-of-the-art performance on several benchmarks by using a single type of descriptors.

Speaker subspace modeling has become increasingly important in speaker recognition, diarization, and clustering. Principal component analysis (PCA) is a popular linear subspace learning technique and the approach that represents an arbitrary utterance or speaker as a linear combination of a set of basis voices based on PCA is known as the eigenvoice approach. In this paper, a novel technique, namely the fishervoice approach, is proposed. The fishervoice approach is based on linear discriminant analysis, another successful linear subspace learning technique that provides an optimized low-dimensional representation of utterances or speakers with focus on the most discriminative basis voices. We apply the fishervoice approach to speaker clustering in a semi-supervised manner and show that the fishervoice approach significantly outperforms the eigenvoice approach in all our experiments on the GALE Mandarin dataset.

In this work, we proposed to use quantitative method to evaluate the accuracy of 3D face reconstruction algorithms. The reconstructed 3D faces are first aligned to the ground truth by iterative closest point (ICP) algorithm and then the shape difference between the two 3D faces is described by signal to noise ratio (SNR). Finally, the error maps (EM) illustrated the reconstruction errors on corresponded vertices in different dimensions. Comparing with the subjective and indirect evaluation methods, the proposed method provides more precise and detailed evaluations for face shape reconstruction. Based on the SNR, different 3D face reconstruction algorithms can be compared directly and the EM also can suggest guidance for feature extraction.

This paper proposes a generative model-based speaker clustering algorithm in the maximum a posteriori adapted Gaussian mixture model (GMM) mean supervector space. The algorithm can be viewed as an extension of the standard expectation maximization algorithm for fitting a mixture model to the data, which iterates between two steps - a sample re-assignment step (E-step) and a model re-estimation step (M-step) - until it converges. The directional scattering patterns of GMM mean supervectors suggest that we employ a mixture of von Mises-Fisher distributions in the model re-estimation step. In the sample re-assignment step, four sample-to-mixture assignment strategies, namely soft, hard, stochastic, and deterministic annealing assignments, are used. Our experiments on the GALE Mandarin dataset show that the use of a mixture of von Mises-Fisher distributions as the underlying model yields significantly higher speaker clustering accuracies than the use of a mixture of Gaussian distributions. It is further shown that deterministic annealing assignment outperforms soft assignment, that soft assignment is comparable to stochastic assignment, and that both soft and stochastic assignments outperform hard assignment.

With the recent efforts made by computer vision researchers, more and more types of features have been designed to describe various aspects of visual characteristics. Modeling such heterogeneous features has become an increasingly critical issue. In this paper, we propose a machinery called the Heterogeneous Feature Machine (HFM) to effectively solve visual recognition tasks in need of multiple types of features. Our HFM builds a kernel logistic regression model based on similarities that combine different features and distance metrics. Different from existing approaches that use a linear weighting scheme to combine different features, HFM does not require the weights to remain the same across different samples, and therefore can effectively handle features of different types with different metrics. To prevent the model from overfitting, we employ the so-called group LASSO constraints to reduce model complexity. In addition, we propose a fast algorithm based on co-ordinate gradient descent to efficiently train a HFM. The power of the proposed scheme is demonstrated across a wide variety of visual recognition tasks including scene, event and action recognition.

In this paper we study some problems related to human age estimation using a large database. First, we study the influence of gender on age estimation based on face representations that combine biologically-inspired features with manifold learning techniques. Second, we study age estimation using smaller gender and age groups rather than on all ages. Significant error reductions are observed in both cases. Based on these results, we designed three frameworks for automatic age estimation that exhibit high performance. Unlike previous methods that require manual separation of males and females prior to age estimation, our work is the first to estimate age automatically on a large database. Furthermore, a data fusion approach is proposed using one of the frameworks, which gives an age estimation error more than 40% smaller than previous methods.

In this paper, we investigate the detection of semantic human actions in complex scenes. Unlike conventional action recognition in well-controlled environments, action detection in complex scenes suffers from cluttered backgrounds, heavy crowds, occluded bodies, and spatial-temporal boundary ambiguities caused by imperfect human detection and tracking. Conventional algorithms are likely to fail with such spatial-temporal ambiguities. In this work, the candidate regions of an action are treated as a bag of instances. Then a novel multiple-instance learning framework, named SMILE-SVM (Simulated annealing Multiple Instance LEarning Support Vector Machines), is presented for learning human action detector based on imprecise action locations. SMILE-SVM is extensively evaluated with satisfactory performances on two tasks: 1) human action detection on a public video action database with cluttered backgrounds, and 2) a real world problem of detecting whether the customers in a shopping mall show an intention to purchase the merchandise on shelf (even if they didn't buy it eventually). In addition, the complementary nature of motion and appearance features in action detection are also validated, demonstrating a boosted performance in our experiments.

Non-frontal view facial expression recognition is important in many scenarios where the frontal view face images may not be available. However, few work on this issue has been done in the past several years because of its technical challenges and the lack of appropriate databases. Recently, a 3D facial expression database (BU-3DFE database) is collected by Yin et al. [10] and has attracted some researchers to study this issue. Based on the BU-3DFE database, in this paper we propose a novel approach to expression recognition from non-frontal view facial images. The novelty of the proposed method lies in recognizing the multi-view expressions under the unified Bayes theoretical framework, where the recognition problem can be formulated as an optimization problem of minimizing an upper bound of Bayes error. We also propose a close-form solution method based on the power iteration approach and rank-one update (ROU) technique to find the optimal solutions of the proposed method. Extensive experiments on BU-3DFE database with 100 subjects and 5 yaw rotation view angles demonstrate the effectiveness of our method.

In this paper, we propose a new image representation to capture both the appearance and spatial information for image classification applications. First, we model the feature vectors, from the whole corpus, from each image and at each individual patch, in a Bayesian hierarchical framework using mixtures of Gaussians. After such a hierarchical Gaussianization, each image is represented by a Gaussian mixture model (GMM) for its appearance, and several Gaussian maps for its spatial layout. Then we extract the appearance information from the GMM parameters, and the spatial information from global and local statistics over Gaussian maps. Finally, we employ a supervised dimension reduction technique called DAP (discriminant attribute projection) to remove noise directions and to further enhance the discriminating power of our representation. We justify that the traditional histogram representation and the spatial pyramid matching are special cases of our hierarchical Gaussianization. We compare our new representation with other approaches in scene classification, object recognition and face recognition, and our performance ranks among the top in all three tasks.

We investigate effective means of building robust dictionaries for detecting the sparse foreground in videos with static background. This work is an extension to our existing solution to foreground/background segmentation problem using the linear programming method proposed to detect sparse errors in signals, which are created by a known dictionary. The dictionary building methods we study are established robust component analysis techniques in the literature (i.e. k-SVD & robust-PCA) as well as a heuristic (running median) inspired by the highly correlated nature of the static video background signal. We compare the effectiveness of the new methods with our original system as well as a baseline method, which is the commonly used single Gaussian model of the background pixels.

Patch-based approaches have become popular in many computer vision applications over recent years. An intrinsic flaw of this framework, missing of the spatial information, however, restricts its usage in face related applications where the spatial configuration is relatively settled. In this paper, we introduce a new patch feature representation, namely Spatial Gaussian Mixture Models (SGMM), which enhances the traditional GMM approach by taking the spatial information into consideration at both local and global scales. In the meantime, SGMM inherits all the merits of GMM, such as precise appearance description and robustness to image misalignment. The experiments on gender recognition demonstrate that the SGMM representation achieves more than 40% relative error reduction compared with either GMM or SVM-based approaches.

This paper demonstrates, to our best knowledge, the first attempt on gender and ethnicity identification from silhouetted face profiles using a computer vision technique. The results achieved, after testing on 441 images, show that silhouetted face profiles have a lot of information, in particular, for ethnicity identification. Shape context based matching was employed for classification. The test samples were multi-ethnic. Average accuracy for gender was 71.20% and for ethnicity 71.66%. However, the accuracy was significantly higher for some classes, such as 83.41% for females (in case of gender identification) and 80.37% for East and South East Asians (in case of ethnicity identification).

In this paper, we propose an efficient speaker clustering approach based on a locality preserving linear projective mapping in the Gaussian mixture model (GMM) mean supervector space. While the GMM mean supervector has turned out to be an effective representation of speakers, its dimensionality is usually very high. The locality preserving projection (LPP) maps the high-dimensional GMM mean supervector space into a lower-dimensional subspace in an unsupervised fashion where the local neighborhood structure of the data points is optimally preserved. Our speaker clustering experiments clearly show that in the reduced-dimensional LPP subspace, traditional clustering techniques such as k-means and hierarchical clustering perform significantly better than they would in the original high-dimensional GMM mean supervector space and in its principal component subspace.

Gaussian mixture models (GMMs) and the minimum error rate classifier (i.e. Bayesian optimal classifier) are popular and effective tools for speech emotion recognition. Typically, GMMs are used to model the class-conditional distributions of acoustic features and their parameters are estimated by the expectation maximization (EM) algorithm based on a training data set. Then, classification is performed to minimize the classification error w.r.t. the estimated class-conditional distributions. We call this method the EM-GMM algorithm. In this paper, we introduce a Boosting algorithm for reliably and accurately estimating the class-conditional GMMs. The resulting algorithm is named the Boosted-GMM algorithm. Our speech emotion recognition experiments show that the emotion recognition rates are effectively and significantly "Boosted" by the Boosted-GMM algorithm as compared to the EM-GMM algorithm. This is due to the fact that the Boosting algorithm can lead to more accurate estimates of the class-conditional GMMs, namely the class-conditional distributions of acoustic features.

Photo community sites such as Flickr and Picasa Web Album host a massive amount of personal photos with millions of new photos uploaded every month. These photos constitute an overwhelming source of images that require effective management. There is an increasingly imperative need for semantic annotation of these web images. This paper addresses the problem by considering two kinds of annotation: semantic annotation and geographic annotation. Both are useful for image search and retrieval and for facilitating communities and social networks. This paper proposes a novel method of Logistic Canonical Correlation Regression (LCCR) for the annotation task. This model exploits the canonical correlation between heterogeneous features and an annotation lexicon of interest, and builds a generalized annotation engine based on canonical correlations in order to produce enhanced annotation for web images. We validate the effectiveness of our algorithm using a dataset of over 380,000 images tagged with GPS coordinates.

In this paper, we propose a new Distributed Video Coding (DVC) algorithm based on Compressive Sampling principles. Our encoding algorithm transmits a set of measurements of every frame block. Using these measurements, the decoder finds an approximation of each block as a linear combination of a small number of blocks in previously transmitted frames. Thanks to the simplicity of the encoding, our algorithm can be useful in those video applications that require very low complex encoders. However, our algorithm is less efficient than another state-of-the-art DVC technique.

The ten papers in this special section focus on applications of probabilistic graphical models in all areas of computer vision.

The success of bilinear subspace learning heavily depends on reducing correlations among features along rows and columns of the data matrices. In this work, we study the problem of rearranging elements within a matrix in order to maximize these correlations so that information redundancy in matrix data can be more extensively removed by existing bilinear subspace learning algorithms. An efficient iterative algorithm is proposed to tackle this essentially integer programming problem. In each step, the matrix structure is refined with a constrained Earth Mover's Distance procedure that incrementally rearranges matrices to become more similar to their low-rank approximations, which have high correlation among features along rows and columns. In addition, we present two extensions of the algorithm for conducting supervised bilinear subspace learning. Experiments in both unsupervised and supervised bilinear subspace learning demonstrate the effectiveness of our proposed algorithms in improving data compression performance and classification accuracy.

Automated analysis of human affective behavior has attracted increasing attention from researchers in psychology, computer science, linguistics, neuroscience, and related disciplines. However, the existing methods typically handle only deliberately displayed and exaggerated expressions of prototypical emotions despite the fact that deliberate behaviour differs in visual appearance, audio profile, and timing from spontaneously occurring behaviour. To address this problem, efforts to develop algorithms that can process naturally occurring human affective behaviour have recently emerged. Moreover, an increasing number of efforts are reported toward multimodal fusion for human affect analysis including audiovisual fusion, linguistic and paralinguistic fusion, and multi-cue visual fusion based on facial expressions, head movements, and body gestures. This paper introduces and surveys these recent advances. We first discuss human emotion perception from a psychological perspective. Next we examine available approaches to solving the problem of machine understanding of human affective behavior, and discuss important issues like the collection and availability of training and test data. We finally outline some of the scientific and engineering challenges to advancing human affect sensing technology.

In this paper, we present a fusion approach to solve the nonrigid shape recovery problem, which takes advantage of both the appearance information and the local features. We have two major contributions. First, we propose a novel progressive finite Newton optimization scheme for the feature-based nonrigid surface detection problem, which is reduced to only solving a set of linear equations. The key is to formulate the nonrigid surface detection as an unconstrained quadratic optimization problem that has a closed-form solution for a given set of observations. Second, we propose a deformable Lucas-Kanade algorithm that triangulates the template image into small patches and constrains the deformation through the second-order derivatives of the mesh vertices. We formulate it into a sparse regularized least squares problem, which is able to reduce the computational cost and the memory requirement. The inverse compositional algorithm is applied to efficiently solve the optimization problem. We have conducted extensive experiments for performance evaluation on various environments, whose promising results show that the proposed algorithm is both efficient and effective.

We propose a five-layer hierarchical space-time model (HSTM) for representing and searching human actions in videos. From a features point of view, both invariance and selectivity are desirable characteristics, which seem to contradict each other. To make these characteristics coexist, we introduce a coarse-to-fine search and verification scheme for action searching, based on the HSTM model. Because going through layers of the hierarchy corresponds to progressively turning the knob between invariance and selectivity, this strategy enables search for human actions ranging from rapid movements of sports to subtle motions of facial expressions. The introduction of the Histogram of Gabor Orientations feature makes the searching for actions go smoothly across the hierarchical layers of the HSTM model. The efficient matching is achieved by applying integral histograms to compute the features in the top two layers. The HSTM model was tested on three selected challenging video sequences and on the KTH human action database. And it achieved improvement over other state-of-the-art algorithms. These promising results validate that the HSTM model is both selective and robust for searching human actions.

This paper introduces a head pose estimation system that automatically localizes the nose-tips of the faces and estimates head poses in images simultaneously. In the training stage, the nose-tips of the faces are first manually labeled. The appearance variations caused by head pose changes are then characterized by a tensorposes model. Given an image with unknown head pose and nose-tip location, the nose-tip of the face is automatically localized in a coarse-to-fine fashion after the skin color segmentation. The head pose is also estimated simultaneously. The performance of our system is evaluated on the Pointing'04 head pose image data set. We first evaluate the classification performance of the tensorposes models with image patches of the faces cropped according to the manually labeled nose-tip locations of the faces in the Pointing '04 data set. By leaving-one-person-out evaluation strategy, we obtain the optimal parameters of the Tensorposes model, and evaluate the discriminative power of the tensorposes model built based on high order singular value decomposition (HOSVD) and multilinear independent component analysis (MICA), and naive principal component analysis (PCA) subspace models. It is shown Tensorposes model by HOSVD and MICA decomposition performs similarly good but much better than naive PCA subspace models. The tensorposes model is then utilized to automatically localize nose-tip location in the testing image and to simultaneously estimate the head pose. The nose-tip localization and pose estimation accuracy of the proposed system are evaluated against the ground truth. Finally, cross-database evaluation of the performance of our system is carried out on Pointing'04 database, a selected subset of CMU PIE database, and some pictures from CLEAR'07 head pose evaluation database. The experiments show that our system generalizes reasonably well to the real-world scenarios.

For the problem of image registration, the top few reliable correspondences are often relatively easy to obtain, while the overall matching accuracy may fall drastically as the desired correspondence number increases. In this paper, we present an efficient feature matching algorithm to employ sparse reliable correspondence priors for piloting the feature matching process. First, the feature geometric relationship within individual image is encoded as a spatial graph, and the pairwise feature similarity is expressed as a bipartite similarity graph between two feature sets; then the geometric neighborhood of the pairwise assignment is represented by a categorical product graph, along which the reliable correspondences are propagated; and finally a closed-form solution for feature matching is deduced by ensuring the feature geometric coherency as well as pairwise feature agreements. Furthermore, our algorithm is naturally applicable for incorporating manual correspondence priors for semi-supervised feature matching. Extensive experiments on both toy examples and real-world applications demonstrate the superiority of our algorithm over the state-of-the-art feature matching techniques.

Precise 3-D head pose estimation plays a significant role in developing human-computer interfaces and practical face recognition systems. This task is challenging due to the particular appearance variations caused by pose changes for a certain subject. In this paper, the pose data space is considered as a union of submanifolds which characterize different subjects, instead of a single continuous manifold as conventionally regarded. A novel manifold embedding algorithm dually supervised by both identity and pose information, called synchronized submanifold embedding (SSE), is proposed for person-independent precise 3-D pose estimation, which means that the testing subject may not appear in the model training stage. First, the submanifold of a certain subject is approximated as a set of simplexes constructed using neighboring samples. Then, these simplexized submanifolds from different subjects are embedded by synchronizing the locally propagated poses within the simplexes and at the same time maximizing the intrasubmanifold variances. Finally, the pose of a new datum is estimated as the propagated pose of the nearest point within the simplex constructed by its nearest neighbors in the dimensionality reduced feature space. The experiments on the 3-D pose estimation database, CHIL data for CLEAR07 evaluation, and the extended application for age estimation on FG-NET aging database, demonstrate the superiority of SSE over conventional regression algorithms as well as unsupervised manifold learning algorithms.

In this corespondence, we study the extra-factor estimation problem with the assumption that the training image ensemble is expressed as an nth-order tensor with the nth-dimension characterizing all features for an image and other dimensions for different extra factors, such as illuminations, poses, and identities. To overcome the local minimum issue of conventional algorithms designed for this problem, we present a novel statistical learning framework called mode-kn Factor Analysis for obtaining a closed-form solution to estimating the extra factors of any test image. In the learning stage, for the kth [see formula in text] dimension of the data tensor, the mode-kn patterns are constructed by concatenating the feature dimension and the kth extra-factor dimension, and then a mode-kn factor analysis model is learnt based on the mode-kn patterns unfolded from the original data tensor. In the inference stage, for a test image, the mode classification of the kth dimension is performed within a probabilistic framework. The advantages of mode-kn factor analysis over conventional tensor analysis algorithms are twofold: 1) a closed-form solution, instead of iterative sub-optimal solution as conventionally, is derived for estimating the extra-factor mode of any test image; and 2) the classification capability is enhanced by interacting with the process of synthesizing data of all other modes in the k th dimension. Experiments on the Pointing'04 and CMU PIE databases for pose and illumination estimation both validate the superiority of the proposed algorithm over conventional algorithms for extra-factor estimation.

In this paper, our contributions to the subspace learning problem are two-fold. We first justify that most popular subspace learning algorithms, unsupervised or supervised, can be unitedly explained as instances of a ubiquitously supervised prototype. They all essentially minimize the intraclass compactness and at the same time maximize the interclass separability, yet with specialized labeling approaches, such as ground truth, self-labeling, neighborhood propagation, and local subspace approximation. Then, enlightened by this ubiquitously supervised philosophy, we present two categories of novel algorithms for subspace learning, namely, misalignment-robust and semi-supervised subspace learning. The first category is tailored to computer vision applications for improving algorithmic robustness to image misalignments, including image translation, rotation and scaling. The second category naturally integrates the label information from both ground truth and other approaches for unsupervised algorithms. Extensive face recognition experiments on the CMU PIE and FRGC ver1.0 databases demonstrate that the misalignment-robust version algorithms consistently bring encouraging accuracy improvements over the counterparts without considering image misalignments, and also show the advantages of semi-supervised subspace learning over only supervised or unsupervised scheme.

Most image annotation systems consider a single photo at a time and label photos individually. In this work, we focus on collections of personal photos and exploit the contextual information naturally implied by the associated GPS and time metadata. First, we employ a constrained clustering method to partition a photo collection into event-based subcollections, considering that the GPS records may be partly missing (a practical issue). We then use conditional random field (CRF) models to exploit the correlation between photos based on 1) time-location constraints and 2) the relationship between collection-level annotation (i.e., events) and image-level annotation (i.e., scenes). With the introduction of such a multilevel annotation hierarchy, our system addresses the problem of annotating consumer photo collections that requires a more hierarchical description of the customers' activities than do the simpler image annotation tasks. The efficacy of the proposed system is validated by extensive evaluation using a sizable geotagged personal photo collection database, which consists of over 100 photo collections and is manually labeled for 12 events and 12 scenes to create ground truth.

Most image annotation systems consider a single photo at a time and label photos individually. In this work, we focus on collections of personal photos and explore the associated GPS and time information for semantic annotation. First, we employ a constrained clustering method to partition a photo collection into event-based sub-collections, considering that the GPS records may be partly missing (a practical issue). We then use conditional random field (CRF) models to exploit the correlation between photos based on (1) time-location constraints and (2) the relationship between collection-level annotation (i.e., events) and image-level annotation (i.e., scenes). With the introduction of such a multi-level annotation hierarchy, our system addresses the problem of annotating consumer photo collections that requires a more hierarchical description of the customers’ activities than do the simpler image annotation tasks. The efficacy of the proposed system is validated using a geotagged customer photo collection database, which consists of over 100 folders and is labeled for 12 events and 12 scenes.

In recent years, local pattern based object detection and recognition have attracted increasing interest in computer vision research community. However, to our best knowledge no previous work has focused on utilizing local patterns for the task of human detection. In this paper we develop a novel human detection system in personal albums based on LBP (local binary pattern) descriptor. Firstly we review the existing gradient based local features widely used in human detection, analyze their limitations and argue that LBP is more discriminative. Secondly, original LBP descriptor does not suit the human detecting problem well due to its high complexity and lack of semantic consistency, thus we propose two variants of LBP: Semantic-LBP and Fourier-LBP. Carefully designed experiments demonstrate the superiority of LBP over other traditional features for human detection. Especially we adopt a random ensemble algorithm for better comparison between different descriptors. All experiments are conducted on INRIA human database.

This paper addresses the problem of recovering 3D human pose from a single monocular image, using a discriminative bag-of-words approach. In previous work, the visual words are learned by unsupervised clustering algorithms. They capture the most common patterns and are good features for coarse-grain recognition tasks like object classification. But for those tasks which deal with subtle differences such as pose estimation, such representation may lack the needed discriminative power. In this paper, we propose to jointly learn the visual words and the pose regressors in a supervised manner. More specifically, we learn an individual distance metric for each visual word to optimize the pose estimation performance. The learned metrics rescale the visual words to suppress unimportant dimensions such as those corresponding to background. Another contribution is that we design an Appearance and Position Context (APC) local descriptor that achieves both selectivity and invariance while requiring no background subtraction. We test our approach on both a quasi-synthetic dataset and a real dataset (HumanEva) to verify its effectiveness. Our approach also achieves fast computational speed thanks to the integral histograms used in APC descriptor extraction and fast inference of pose regressors.

In this paper, we study the problem of subspace-based face recognition under scenarios with spatial misalignments and/or image occlusions. For a given subspace, the embedding of a new datum and the underlying spatial misalignment parameters are simultaneously inferred by solving a constrained ℓ1norm optimization problem, which minimizes the error between the misalignment-amended image and the image reconstructed from the given subspace along with its principal complementary subspace. A byproduct of this formulation is the capability to detect the underlying image occlusions. Extensive experiments on spatial misalignment estimation, image occlusion detection, and face recognition with spatial misalignments and image occlusions all validate the effectiveness of our proposed general formulation.

Background estimation, i.e. automatic recovery of the background image from a sequence of images containing moving foreground objects, is an important module in many applications, e.g. surveillance and video segmentation. In this paper, we present a simple, yet effective and robust approach for background estimation based on Loopy Belief Propagation. Robustness of the proposed approach means: (i) minimal assumption on the input frames, and (ii) no need to tune parameters. Basically, the background can be recovered even when the occluding foreground objects stay still for a long time. Furthermore, no motion information needs to be known or estimated for the foreground objects, which implies that background can be recovered from a set of frames which are not consecutive temporally. Analysis and experiments are provided to compare the proposed approach to related methods. Experimental results on typical surveillance videos demonstrate the effectiveness of our approach.

In this paper, we present a patch-based regression framework for addressing the human age and head pose estimation problems. Firstly, each image is encoded as an ensemble of orderless coordinate patches, the global distribution of which is described by Gaussian Mixture Models (GMM), and then each image is further expressed as a specific distribution model by Maximum a Posteriori adaptation from the global GMM. Then the patch-kernel is designed for characterizing the Kullback-Leibler divergence between the derived models for any two images, and its discriminating power is further enhanced by a weak learning process, called inter-modality similarity synchronization. Finally, kernel regression is employed for ultimate human age or head pose estimation. These three stages are complementary to each other, and jointly minimize the regression error. The effectiveness of this regression framework is validated by three experiments: 1) on the YAMAHA aging database, our solution brings a more than 50% reduction in age estimation error compared with the best reported results; 2) on the FG-NET aging database, our solution based on raw image features performs even better than the state-of-the-art algorithms which require fine face alignment for extracting warped appearance features; and 3) on the CHIL head pose database, our solution significantly outperforms the best one reported in the CLEAR07 evaluation.

This paper addresses the problem of generating a super-resolution (SR) image from a single low-resolution input image. We approach this problem from the perspective of compressed sensing. The low-resolution image is viewed as downsampled version of a high-resolution image, whose patches are assumed to have a sparse representation with respect to an over-complete dictionary of prototype signal-atoms. The principle of compressed sensing ensures that under mild conditions, the sparse representation can be correctly recovered from the downsampled signal. We will demonstrate the effectiveness of sparsity as a prior for regularizing the otherwise ill-posed super-resolution problem. We further show that a small set of randomly chosen raw patches from training images of similar statistical nature to the input image generally serve as a good dictionary, in the sense that the computed representation is sparse and the recovered high-resolution image is competitive or even superior in quality to images produced by other SR methods.

We introduce a general formulation, called non-negative graph embedding, for non-negative data decomposition by integrating the characteristics of both intrinsic and penalty graphs [17]. In the past, such a decomposition was obtained mostly in an unsupervised manner, such as Non-negative Matrix Factorization (NMF) and its variants, and hence unnecessary to be powerful at classification. In this work, the non-negative data decomposition is studied in a unified way applicable for both unsupervised and supervised/semi-supervised configurations. The ultimate data decomposition is separated into two parts, which separatively preserve the similarities measured by the intrinsic and penalty graphs, and together minimize the data reconstruction error. An iterative procedure is derived for such a purpose, and the algorithmic non-negativity is guaranteed by the non-negative property of the inverse of any M-matrix. Extensive experiments compared with NMF and conventional solutions for graph embedding demonstrate the algorithmic properties in sparsity, classification power, and robustness to image occlusions.

In this paper, we address the pair-activity classification problem, which explores the relationship between two active objects based on their motion information. Our contributions are three-fold. First, we design a set of features, e.g., causality ratio and feedback ratio based on the Granger Causality Test (GCT), for describing the pair-activities encoded as trajectory pairs. These features along with conventional velocity and position features are essentially of multi-modalities, and may be greatly different in scale and importance. To make full use of them, we then present a novel feature normalization procedure to learn the coefficients for weighting these features by maximizing the discriminating power measured by weighted correlation. Finally, we collected a pair-activity database of five categories, each of which consists of about 170 instances. The extensive experiments on this database validate the effectiveness of the designed features for pair-activity representation, and also demonstrate that the proposed feature normalization procedure greatly boosts the pair-activity classification accuracy.

The ability to handle multi-view facial expressions is important for computers to understand affective behavior under less constrained environment. However, most of existing methods for facial expression recognition are based on the near-frontal view face data, which are likely to fail in the non-frontal facial expression analysis. In this paper, we conduct an investigation on analyzing multi-view facial expressions. Three local patch descriptors (HoG, LBP, and SIFT) are used to extract facial features, which are the inputs to a nearest-neighbor indexing method that identifies facial expressions. We also investigate the influence of feature dimension reductions (PCA, LDA, and LPP) and classifier fusion on the recognition performance. We test our approaches on multi-view data generated from BU-3DFE 3D facial expression database that includes 100 subjects with 6 emotions and 4 intensity levels. Our extensive person-independent experiments suggest that the SIFT descriptor outperforms HoG and LBP, and LPP outperforms PCA and LDA in this application. But the classifier fusion does not show a significant advantage over SIFT-only classifier.

The 3D facial geometry contains ample information about human facial expressions. Such information is invariant to pose and lighting conditions, which have imposed serious hurdles on many 2D facial analysis problems. In this paper, we perform person and gender independent facial expression recognition based on properties of the line segments connecting certain 3D facial feature points. The normalized distances and slopes of these line segments comprise a set of 96 distinguishing features for recognizing six universal facial expressions, namely anger, disgust, fear, happiness, sadness, and surprise. Using a multi-class support vector machine (SVM) classifier, an 87.1% average recognition rate is achieved on the publicly available 3D facial expression database BU-3DFE [1]. The highest average recognition rate obtained in our experiments is 99.2% for the recognition of surprise. Our result outperforms the result reported in the prior work [2], which uses elaborately extracted primitive facial surface features and an LDA classifier and which yields an average recognition rate of 83.6% on the same database.

Image representation and distance metric are both significant for learning-based visual classification. This paper presents the concept of k-Nearest-Neighbor Simplex (kNNS), which is a simplex with the vertices as the k nearest neighbors of a certain point. kNNS contributes to the image classification problem in two aspects. First, a novel distance metric between a point to its kNNS within a certain class is provided for general classification problem. Second, we develop a new subspace learning algorithm, called Discriminant Simplex Analysis (DSA), to pursue effective feature representation for image classification. In DSA, the within-locality and between-locality are both modeled by kNNS distance, which provides a more accurate and robust measurement of the probability of a point belonging to a certain class. Experiments on real-world image classification demonstrate the effectiveness of both DSA as well as kNNS based classification approach.

This paper proposes a novel camera/microphone array system capable of capturing dynamic facial expression video with synchronized speech and reconstructing realistic 3D face models from the data. Both hardware and software issues including camera calibration, video/audio synchronization, facial marker tracking and 3D shape reconstruction are considered. To our best knowledge, this system is the first camera/microphone array system that is able to capture high-resolution facial expression video with synchronized speech. The system can be used to collect dynamic 3D audiovisual face data for many multimedia applications.

In this paper we propose a framework of factorization-based non-rigid shape modeling and tracking in stereo-motion. We construct a measurement matrix with the stereo-motion data captured from a stereo-rig. Organized in a particular way this matrix could be decomposed by Singular Value Decomposition (SVD) into the 3D basis shapes, their configuration weights, rigid motion and camera geometry. Accordingly, the stereo correspondences can be inferred from motion correspondences only requiring that a minimum of 3K point stereo correspondences (where K is the dimension of shape basis space) are created in advance. Basically this framework still keeps the property of rank constraints, meanwhile it owns other advantages such as simpler correspondence and accurate reconstruction even with short image sequences. Results with real data are given to demonstrate its performance.

By combining audio and visual modalities, the speech recognition systems achieve higher performance and robustness. The fusion strategies to this point are mainly three types: feature level fusion model level fusion and decision level fusion. In this paper, we present a novel audio-visual fusion framework, in which a joint dimensionality reductional approach is used to project the audio and visual features into more compact subspaces. With correlation preserving criteria, the representations of projected audio and visual features will be able to preserve the correlation conveyed in the original audio and visual feattre space. At the same time, the better model efficiency is achieved in the more compact feature spaces. The experiments on audio-visual person verification demonstrate the efficiency and effectiveness of the proposed fusion framework.

In this paper, illuminated by the great success of Universal Background Modeling (UBM) for speech/speaker recognition, we present a new algorithm for face recognition. On the one hand, we encode each face image as an ensemble of X-Y patches, which integrate both local appearance and shape information. These X-Y patch representation provides the possibility to compare two spatially different patches, and consequently alleviates the requirement of exact pixel-wise alignment. On the other hand, we train the UBM based on the X-Y patches from the images of different subjects, and then automatically adapt the UBM for specific subject, and finally face recognition is conducted by comparing the ratio of the likelihoods from the model for specific subject and UBM. UBM elicits the algorithmic robustness to image occlusion since the occluded patches may not contribute evidence to any subjects. Comparison experiments with the state-of-the-art subspace learning algorithms, on the popular CMU PIE face database and with varieties of configurations, demonstrate that our proposed algorithm brings significant improvement in face recognition accuracy, and also show the algorithmic robustness to image occlusions.

Motivated by the fact that age information can often be observed from local evidence on the human face, we contribute to the age estimation problem in two aspects. On the one hand, we present a new feature descriptor, called spatially flexible patch (SFP), which encodes the local appearance and position information simultaneously. SFP has the potential to alleviate the problem of insufficient samples owing to that SFPs similar in appearance yet slightly different in position can still provide similar confidence for age estimation. One the other hand, the SFP associated with age label is modeled with Gaussian Mixture Model, and then age estimation is conducted by maximizing the sum of likelihoods from all the SFPs associated with the hypothetic age. Experiments are conducted on the YAMAHA database with 8,000 face images and ages ranging from 0 to 93. Compared with the latest reported results, our new algorithm brings encouraging reduction in mean absolute error for age estimation.

It is a fascinating yet challenging problem to accurately and efficiently localize regionally distinct features between face groups in multi-dimensional signal processing and analysis. Given a data with unknown distribution and small sample size, we propose a new statistical analysis framework using hybrid randomization (i.e., permutation) tests to improve the system's efficiency in identifying distinct features. The proposed method fits the nonparametric distribution of the test statistic with Pearson distribution series. We bypass the tedious online randomization via calculating the first four moments of the permutation distribution. This can reduce the computational complexity from O(n!) to O(n2) over traditional methods for the modified Hotelling's T2test statistics. Experiments on simulated data and 3D face analysis demonstrate the efficiency, accuracy and robustness of the proposed approach.

Gaussian mixture models (QMM) have become one of the standard acoustic approaches for Language Detection. These models are typically incorporated to produce a log-likelihood ratio (LLR) verification statistic. In this framework, the intersession variability within each language becomes an adverse factor degrading the accuracy. To address this problem, we formulate the LLR as a function of the QMM parameters concatenated into normalized mean supervectors, and estimate the distribution of each language in this (high dimensional) supervector space. The goal is to de-emphasize the directions with the largest intersession variability. We compare this method with two other popular intersession variability compensation methods known as Nuisance Attribute Projection (NAP) and Within-Class Covariance Normalization (WCCN). Experiments on the NIST LRE 2003 and NIST LRE 2005 speech corpora show that the presented technique reduces the error by 50% relative to the baseline, and performs competitively with the NAP and WCCN approaches. Fusion results with a phonotactic component are also presented.

Speech perceptual features, such as Mel-frequency Cepstral Coefficients (MFCC), have been widely used in acoustic event detection. However, the different spectral structures between speech and acoustic events degrade the performance of the speech feature sets. We propose quantifying the discriminative capability of each feature component according to the approximated Bayesian accuracy and deriving a discriminative feature set for acoustic event detection. Compared to MFCC, feature sets derived using the proposed approaches achieve about 30% relative accuracy improvement in acoustic event detection.

To design geometrically motivated approaches for classifying the high-dimensional data, we propose to learn a discriminant subspace using Correlation Embedding Analysis (CEA). This novel algorithm enhances its discriminant power by incorporating both correlational graph embedding and Fisher criterion. In a geometric interpretation, it projects the high- dimensional data onto a hypersphere and preserves intrinsic neighbor relations with the Pearson correlation metric. After the embedding, resulting data pairs from the same class are forced to enhance their correlation affinity, whereas neighboring points of different class are forced to reduce their correlation affinity at the same time. The feature learned by CEA is tolerable to scaling or outlier. Experiments on face recognition demonstrate the effectiveness and advantage of the CEA.

This paper addresses the problem of recovering 3D human pose from a single monocular image. In the literature, Bayesian Mixtures of Experts (BME) was successfully used to represent the multimodal image-to-pose distributions. However, the expectation-maximization (EM) algorithm that learns the BME model may converge to a suboptimal local maximum. And the quality of the final solution depends largely on the initial values. In this paper, we propose an efficient initialization method for BME learning. We first partition the training set so that each subset can be well modeled by a single expert and the total regression error is minimized. Then each expert and gate of BME model is initialized on a partition subset. Our initialization method is tested on both a quasi-synthetic dataset and a real dataset (HumanEva). Results show that it greatly reduces the computational cost in training while improves testing accuracy.

In this paper, an MPEG4 real-time performance-driven avatar is proposed. The facial motion parameters, namely head rotation angles and translation distances as well as a set of action unit weights, are estimated from live video on a per-frame basis using a two-step robust facial motion tracking algorithm. The estimated parameters are then converted into MPEG4 facial animation parameter (FAP) values by solving an overdetermined constrained linear least-squares optimization problem. These FAP values can be used to drive an MPEG4-compliant face model to generate on-the-fly realistic facial animation that resembles the facial actions of the performer in the video. Such a real-time performance-driven avatar has important applications in a broad range of areas including very low bit rate video communication, movie and game industries as well as human-computer intelligent interaction.

In this paper, we address the problem of hallucinating a high resolution face given a low resolution input face. The problem is approached through sparse coding. To exploit the facial structure, non-negative matrix factorization (NMF) is first employed to learn a localized part-based subspace. This subspace is effective for super-resolving the incoming low resolution face under reconstruction constraints. To further enhance the detailed facial information, we propose a local patch method based on sparse representation with respect to coupled overcomplete patch dictionaries, which can be fast solved through linear programming. Experiments demonstrate that our approach can hallucinate high quality super-resolution faces.

We propose a novel method to represent motion trajectory - ‘bag of segments’. Motivated by the ‘bag of words’ approach in text mining and more recently in image categorization, we represent each trajectory as a collection of segments, each assigned a membership to a codeword of a dictionary. The trajectory segments are represented by their shape and motion information while the inter-segment relationship is ignored. In the ’bag of trajectory segments’ representation, the trajectories are transformed to a vector in the codeword space. The codebook is generated by clustering a large amount of trajectory segments using Expectation Maximization (EM) algorithm and each codeword is a weighted regression model. Given the codebook, trajectories are segmented and associated with a soft-frequency vector in codeword space as an intermediate representation. With the ‘bag of segments’ representation, trajectory analysis can be carried out in vector space as traditional data. Experiments show the ‘bag of segments’ method is very effective in trajectory similarity search and classification.

This paper proposes a fully automatic framework for static human head pose estimation. With a 2D human multi-view face image as input, the face region is detected and cropped out. Then the pose of the face is assessed by the pose categories. Based on the appearance of the face region, variant subspace learning methods including Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), Locality Preserving Projection (LPP) and Pose-Specific Subspace (PSS) are proposed for effective representation of the face poses. Several aspects, such as human identification, illumination changes and expression variations are considered during the classification process. The experiment results on large public database demonstrate the effectiveness of the proposed framework and recognition algorithms. Performance comparisons and discussions are also provided in detail to help the algorithm selection when designing practical face pose estimation systems for different scenarios.

In this paper, we propose a complete pipeline of efficient and low-cost techniques to construct a realistic 3D text-driven emotive audio-visual avatar from a single 2D frontal-view face image of any person on the fly. This real-time conversion is achieved through three steps. First, a personalized 3D face model is built based on the 2D face image using a fully automatic 3D face shape and texture reconstruction framework. Second, using standard MPEG-4 FAPs (Facial Animation Parameters), the face model is animated by the viseme and expression channels and is complemented by the visual prosody channel that controls head, eye and eyelid movements. Finally, the facial animation is combined and synchronized with the emotive synthetic speech generated by incorporating an emotion transformer into a Festival-MBROLA text to neutral speech synthesizer.

Frames of videos with static background and dynamic foreground can be viewed as samples of signals that vary slowly in time with sparse corruption caused by foreground objects. We cast background subtraction as a signal estimation problem, where the error sparsity is enforced through minimization of the L1norm of the difference between the processed frame and estimated background subspace, as an approximation to the underlying L0norm minimization structure. Our work provides a novel framework for background subtraction with the added benefit of easy integration of local discriminative information (e.g. gradient, texture, motion field etc.) for improved robustness. We show that the proposed method is able to overcome various difficulties frequently encountered in real application settings, and is competitive with the state of the art.

Head pose estimation has many useful applications in practice. How to estimate the head pose automatically and robustly is still a challenging problem. In pose estimation, different pose angles can be used as regression values or viewed as different class labels. Thus a question is raised in our study: which is proper for pose estimation - classification or regression? We investigate representative classification and regression methods on the same problem to see any difference. A method that combines regression and classification approaches is also examined. Preliminary experiments show some interesting results which might prompt further exploration of related issues in pose estimation.

In this work we investigate the relationship between Bregman distances and regularized logistic regression model. We convert L1-regularized logistic regression (LR) into more general Bregman divergence framework and propose a primal-dual method based algorithm for learning the parameters of the model. The proposed method utilizes L1 regularization to incorporate parameter sparsity into the divergence minimization scheme. We perform tests on public domain data sets and produce results which are amongst the best reported.

The existing methods of facial expression recognition are typically based on the near-frontal face data. The analysis of non-frontal-view facial expression is a largely unexplored research. The accessibility to a recent 3D facial expression database (BU-3DFE database) motivates us to explore an interesting question: whether non-frontal-view facial expression analysis can achieve the same as or better performance than the existing frontal-view facial expression method. Our extensive recognition experiments on data of 100 subjects with 5 yaw rotation view angles suggests that the non-frontal-view facial expression classification can outperform frontal-view facial expression recognition, given the manually labeled facial key points.

This paper proposes a general framework to detect unsafe states of a system whose basic realtime parameters are captured by multi-sensors. Our approach is to learn a danger level function which can be used to alert the users in advance of dangerous situations. The main challenge to this learning problem is the labelling issue, i.e., it is difficult to assign an objective danger level at each time step to the training data, except at the collapse points where a penalty can be assigned and at the successful ends where a certain reward can be assigned. In this paper, we treat the danger level as expected future reward (penalty is regarded as negative reward) and use temporal difference (TD) learning [2] to learn a function to approximate the expected future reward. The TD learning obtains the approximation by propagating the penalty/reward observable at collapse points or successful ends to the entire feature space following some constraints. Our approach is applied to, but not limited to, the application of monitoring of driving safety and the experimental results demonstrate the effectiveness of the approach.

The Gaussian mixture model (GMM) can approximate arbitrary probability distributions, which makes it a powerful tool for feature representation and classification. However, it suffers from insufficient training data, especially when the feature space is of high dimensionality. In this paper, we present a novel approach to boost the GMMs via discriminant analysis in which the required amount of training data depends only upon the number of classes, regardless of the feature dimension. We demonstrate the effectiveness of the proposed BoostGMM-DA classifier by applying it to the problem of emotion recognition in speech. Our experiment results indicate that significantly higher recognition rates are achieved by the BoostGMM-DA classifier than are achieved by the conventional GMM minimum error rate (MER) classifier under the same training conditions, and that significantly less training data are required for the BoostGMM-DA classifier to yield comparable recognition rates to the GMM MER classifier.

In this paper, we present an efficient discriminative method for human pose estimation. This method learns a direct mapping from visual observations to human body configurations. The framework requires that the visual features should be powerful enough to discriminate the subtle differences between similar human poses. We propose to describe the image features using salient interest points that are represented by SIFT-like descriptors. The descriptor encode the position, appearance, and local structural information simultaneously. Bag-of-words representation is used to model the distribution of feature space. The descriptor can tolerate a range of illumination and position variations because it is computed on overlapped patches. We use Gaussian process regression to model the mapping from visual observations to human poses. This probabilistic regression algorithm is effective and robust to the pose estimation problem. We test our approach on the HumanEva data set. Experimental results demonstrate that our approach achieves the state of the art performance.

This paper presents a novel Gaussianized vector representation for scene images by an unsupervised approach. First, each image is encoded as an ensemble of orderless bag of features, and then a global Gaussian Mixture Model (GMM) learned from all images is used to randomly distribute each feature into one Gaussian component by a multinomial trial. The parameters of the multinomial distribution are defined by the posteriors of the feature on all the Gaussian components. Finally, the normalized means of the features distributed in every Gaussian component are concatenated to form a supervector, which is a compact representation for each scene image. We prove that these super-vectors observe the standard normal distribution. Our experiments on scene categorization tasks using this vector representation show significantly improved performance compared with the bag-of-features representation.

Recent studies in patch-based Gaussian Mixture Model (GMM) approaches for face age estimation present promising results. We propose using a hidden Markov model (HMM) supervector to represent face image patches, to improve from the previous GMM supervector approach by capturing the spatial structure of human faces and loosening the assumption of identical face patch distribution within a face image. The Euclidean distance of HMM supervectors constructed from two face images measures the similarity of the human faces, derived from the approximated Kullback-Leibler divergence between the joint distributions of patches with implicit unsupervised alignment of different regions in two human faces. The proposed HMM supervector approach compares favorably with the GMM supervector approach in face age estimation on a large face dataset.

This paper describes a database of dysarthric speech produced by 19 speakers with cerebral palsy. Speech materials consist of 765 isolated words per speaker: 300 distinct uncommon words and 3 repetitions of digits, computer commands, radio alphabet and common words. Data is recorded through an 8-microphone array and one digital video camera. Our database provides a fundamental resource for automatic speech recognition development for people with neuromotor disability. Research on articulation errors in dysarthria will benefit clinical treatments and contribute to our knowledge of neuromotor mechanisms in speech production. Data files are available via secure ftp upon request. Index Terms: speech recognition, dysarthria, cerebral palsy 1.

In this paper, we adopt a difference approach to prosody prediction for emotional text-to-speech synthesis, where the prosodic variations between emotional and neutral speech are decomposed into the global and local prosodic variations and predicted using a two-stage model. The global prosodic variations are modeled by the means and standard deviations of the prosodic parameters, while the local prosodic variations are modeled by the classification and regression tree (CART) and dynamic programming. The proposed two-stage prosody prediction model has been successfully implemented as a prosodic module in a Festival-MBROLA architecture based emotional text-to-speech synthesis system, which is able to synthesize highly intelligible, natural and expressive speech. Index Terms: TTS, speech synthesis, prosody prediction, CART, dynamic programming

This paper studies the problem of recognizing gender from full body images. This problem has not been addressed before, partly because of the variant nature of human bodies and clothing that can bring tough difficulties. However, gender recognition has high application potentials, e.g. security surveillance and customer statistics collection in restaurants, supermarkets, and even building entrances. In this paper, we build a system of recognizing gender from full body images, taken from frontal or back views. Our contributions are three-fold. First, to handle the variety of human body characteristics, we represent each image by a collection of patch features, which model different body parts and provide a set of clues for gender recognition. To combine the clues, we build an ensemble learning algorithm from those body parts to recognize gender from fixed view body images (frontal or back). Second, we relax the fixed view constraint and show the possibility to train a flexible classifier for mixed view images with the almost same accuracy as the fixed view case. At last, our approach is shown to be robust to small alignment errors, which is preferred in many applications.

This paper considers the emerging problem of annotating personal photo collections that are taken by digital cameras and may have been subsequently organized by customers. Unlike the images from the web searching engine or commercial image banks (e.g. the Corel database), the photos in the same personal collection are related to each other in time, location, and content. Advanced technologies can record the GPS coordinates for each photo, and thus provide a richer source of context to model and enforce the correlation between the photos in the same collection. Recognizing the well-known limitations ("semantic gap") of visual recognition algorithms, we exploit the correlation between the photos to enhance the annotation performance. In our approach, high-confidence annotation labels are first obtained for certain photos and then propagated to the remaining photos in the same collection, according to time, location, and visual proximity (or similarity). A novel generative probabilistic model is employed, which outperforms the pervious linear propagation scheme. Experimental results have shown the advantages of the proposed annotation scheme.

The explosive growth of video content in recent years fueled by the technological leaps in computing and communication has created new challenges for video content analysis that can serve applications in video surveillance, video searching and mining. Human action detection and recognition is one of the important tasks in this effort. In this paper, we present a luminance field manifold trajectory analysis based solution for human activity recognition, without explicit object level information extraction and understanding. This approach is computationally efficient and can operate in real time. The recognition performance is also comparable with the state of art in comparable set ups.

In this work, we present a SIFT-Bag based generative-todiscriminative framework for addressing the problem of video event recognition in unconstrained news videos. In the generative stage, each video clip is encoded as a bag of SIFT feature vectors, the distribution of which is described by a Gaussian Mixture Models (GMM). In the discriminative stage, the SIFT-Bag Kernel is designed for characterizing the property of Kullback-Leibler divergence between the specialized GMMs of any two video clips, and then this kernel is utilized for supervised learning in two ways. On one hand, this kernel is further refined in discriminating power for centroid-based video event classification by using the Within-Class Covariance Normalization approach, which depresses the kernel components with high-variability for video clips of the same event. On the other hand, the SIFT-Bag Kernel is used in a Support Vector Machine for margin-based video event classification. Finally, the outputs from these two classifiers are fused together for final decision. The experiments on the TRECVID 2005 corpus demonstrate that the mean average precision is boosted from the best reported 38.2 % in [36] to 60.4 % based on our new framework.

Automatic human age estimation has considerable potential applications in human computer interaction and multimedia communication. However, the age estimation problem is challenging. We design a locally adjusted robust regressor (LARR) for learning and prediction of human ages. The novel approach reduces the age estimation errors significantly over all previous methods. Experiments on two aging databases show the success of the proposed method for human aging estimation.

Emotive audio-visual avatars have the potential of significantly improving the quality of Human-Computer Interaction (HCI). In this paper, the various technical approaches of a novel framework leading to a text-driven 3D Emotive Audio-Visual Avatar (EAVA) are proposed. Primary work is focused on 3D face modeling, realistic emotional facial expression animation, emotive speech synthesis, and the co-articulation of speech gestures (i.e., lip movements due to speech production) and facial expressions. Experimental results clearly indicate that a certain degree of naturalness and expressiveness has been achieved by EAVA in both audio and visual aspects. Promising potential improvements can be expected by incorporating various data-driven statistical learning models into the framework.

In many video-based object recognition applications, the object appearances are acquired by visual tracking or detection and are inconsistent due to misalignments. We believe the misalignments can be removed if we can reduce the inconsistency in the object appearances caused by misalignments through clustering the objects in appearance, space and time domain simultaneously. We therefore propose to learn Transform Invariant Mixtures of Probabilistic PCA (TIMPPCA) model from the data while at the same time eliminating the misalignments. The model is formulated in a generative framework, and the misalignments are considered as hidden variables in the model. Variational EM update rules are then derived based on Variational Message Passing (VMP) techniques. The proposed TIMP-PCA is applied to improve head pose estimation performance and to detect the change of attention focus in meeting room video for meeting room video indexing/retrieval and achieves promising performance.

In a higher level task such as clustering of web results or

Beyond conventional linear and kernel-based feature extraction, we propose in this paper the generalized feature extraction formulation based on the so-called Graph Embedding framework. Two novel correlation metric based algorithms are presented based on this formulation. Correlation Embedding Analysis (CEA), which incorporates both correlational mapping and discriminating analysis, boosts the discriminating power by mapping data from a high-dimensional hypersphere onto another low-dimensional hypersphere and preserving the intrinsic neighbor relations with local graph modeling. Correlational Principal Component Analysis (CPCA) generalizes the conventional Principal Component Analysis (PCA) algorithm to the case with data distributed on a high-dimensional hypersphere. Their advantages stem from two facts: 1) tailored to normalized data, which are often the outputs from the data preprocessing step, and 2) directly designed with correlation metric, which shows to be generally better than Euclidean distance for classification purpose. Extensive comparisons with existing algorithms on visual classification experiments demonstrate the effectiveness of the proposed algorithms.

As the first decade of the 21st century comes to a close, growth in multimedia delivery infrastructure and public demand for applications built on this backbone are converging like never before. The push towards reaching truly interactive multimedia technologies becomes stronger as our media consumption paradigms continue to change. In this paper, we profile a technology leading the way in this revolution: active learning. Active learning is a strategy that helps alleviate challenges inherent in multimedia information retrieval through user interaction. We show how active learning is ideally suited for the multimedia information retrieval problem by giving an overview of the paradigm and component technologies used with special attention given to the application scenarios in which these technologies are useful. Finally, we give insight into the future of this growing field and how it fits into the larger context of multimedia information retrieval.

This paper presents a novel real-time multimodal human-avatar interaction (RTM-HAI) framework with vision-based remote animation control (RAC). The framework is designed for both mobile and desktop avatar-based human-machine or human-human visual communications in real-world scenarios. Using 3-D components stored in the Java mobile 3-D (M3G) file format, the avatar models can be flexibly constructed and customized on the fly on any mobile devices or systems that support the M3G standard. For the RAC head tracker, we propose a 2-D real-time face detection/tracking strategy through an interactive loop, in which the detection and tracking complement each other for efficient and reliable face localization, tolerating extreme user movement. With the face location robustly tracked, the RAC head tracker selects a main user and estimates the user's head rolling, tilting, yawing, scaling, horizontal, and vertical motion in order to generate avatar animation parameters. The animation parameters can be used either locally or remotely and can be transmitted through socket over the network. In addition, it integrates audio-visual analysis and synthesis modules to realize multichannel and runtime animations, visual TTS and real-time viseme detection and rendering. The framework is recognized as an effective design for future realistic industrial products of humanoid kiosk and human-to-human mobile communication.

Conventional subspace learning or recent feature extraction methods consider globality as the key criterion to design discriminative algorithms for image classification. We demonstrate in this paper that applying the local manner in sample space, feature space, and learning space via linear subspace learning can sufficiently boost the discriminating power, as measured by discriminating power coefficient (DPC). The proposed solution achieves good classification accuracy gains and shows computationally efficient. Particularly, we approximate the global nonlinearity through a multimodal localized piecewise subspace learning framework, in which three locality criteria can work individually or jointly for any new subspace learning algorithm design. It turns out that most existing subspace learning methods can be unified in such a common framework embodying either the global or local learning manner. On the other hand, we address the problem of numerical difficulty in the large-size pattern classification case, where many local variations cannot be adequately handled by a single global model. By localizing the modeling, the classification error rate estimation is also localized and thus it appears to be more robust and flexible for the model selection among different model candidates. As a new algorithm design based on the proposed framework, the query-driven locally adaptive (QDLA) mixture-of-experts model for robust face recognition and head pose estimation is presented. Experiments demonstrate the local approach to be effective, robust, and fast for large size, multiclass, and multivariance data sets.

Recent research has demonstrated the success of supervised dimensionality reduction algorithms 2DLDA and 2DMFA, which are based on the image-as-matrix representation, in small sample size cases. To solve the convergence problem in 2DLDA and 2DMFA, we propose in this work two new schemes, called Null Space based 2DLDA (NS2DLDA) and Null Space based 2DMFA (NS2DMFA), and apply them to the challenging multi-view face recognition task. First, we convert each 2-D face image (matrix) into a vector and compute the first projection matrixP1from the null space of the intra-class scatter matrix, such that the samples from the same class are projected to the same point. Then the data are projected and reconstructed withP1. Finally, we re-organize the reconstructed datum into a matrix and then compute the second projection directionP2, in the form of a Kronecker product of two matrices, by maximizing the inter-class scatter. A proof of algorithmic convergence is provided. The experiments on two benchmark multi-view face databases, the CMU PIE and FERET databases, demonstrate that NS2DLDA outperforms Fisherface, Null Space LDA (NSLDA) and 2DLDA. Additionally, NS2DMFA is also demonstrated to be more accurate than MFA and 2DMFA for face recognition.

Principal Components Analysis (PCA) has traditionally been utilized with data expressed in the form of 1-D vectors, but there exists much data such as gray-level images, video sequences, Gabor-filtered images and so on, that are intrinsically in the form of second or higher order tensors. For representations of image objects in their intrinsic form and order rather than concatenating all the object data into a single vector, we propose in this paper a new optimal object reconstruction criterion with which the information of a high-dimensional tensor is represented as a much lower dimensional tensor computed from projections to multiple concurrent subspaces. In each of these subspaces, correlations with respect to one of the tensor dimensions are reduced, enabling better object reconstruction performance. Concurrent subspaces analysis (CSA) is presented to efficiently learn these subspaces in an iterative manner. In contrast to techniques such as PCA which vectorize tensor data, CSA's direct use of data in tensor form brings an enhanced ability to learn a representative subspace and an increased number of available projection directions. These properties enable CSA to outperform traditional algorithms in the common case of small sample sizes, where CSA can be effective even with only a single sample per class. Extensive experiments on images of faces and digital numbers encoded as second or third order tensors demonstrate that the proposed CSA outperforms PCA-based algorithms in object reconstruction and object recognition.

Techniques for classification and feature extraction are often intertwined. In this paper, we contribute to these two aspects via the shared philosophy of simplexizing the sample set. For general classification, we present a new criteria based on the concept of -nearest-neighbor simplex (), which is constructed by the nearest neighbors, to determine the class label of a new datum. For feature extraction, we develop a novel subspace learning algorithm, called discriminant simplex analysis (DSA), in which the intraclass compactness and interclass separability are both measured by distances. Comprehensive experiments on face recognition and lipreading validate the effectiveness of the DSA as well as the -based classification approach.

In this paper, we investigate two soft-biometric problems: (1) age estimation and (2) pose estimation, within the scenario where uncertainties exist for the available labels of the training samples. These two tasks are generally formulated as the automatic design of a regressor from training samples with uncertain nonnegative labels. First, the nonnegative label is predicted as the Frobenius norm of a matrix, which is bilinearly transformed from the nonlinear mappings of a set of candidate kernels. Two transformation matrices are then learned for deriving such a matrix by solving two semidefinite programming (SDP) problems, in which the uncertain label of each sample is expressed as two inequality constraints. The objective function of SDP controls the ranks of these two matrices and, consequently, automatically determines the structure of the regressor. The whole framework for the automatic design of a regressor from samples with uncertain nonnegative labels has the following characteristics: (1) the SDP formulation makes full use of the uncertain labels, instead of using conventional fixed labels; (2) regression with the Frobenius norm of matrix naturally guarantees the nonnegativity of the labels, and greater prediction capability is achieved by integrating the squares of the matrix elements, which to some extent act as weak regressors; and (3) the regressor structure is automatically determined by the pursuit of simplicity, which potentially promotes the algorithmic generalization capability. Extensive experiments on two human age databases: (1) FG-NET and (2) Yamaha, and the Pointing'04 head pose database, demonstrate encouraging estimation accuracy improvements over conventional regression algorithms without taking the uncertainties within the labels into account.

Images, as high-dimensional data, usually embody large variabilities. To classify images for versatile applications, an effective algorithm is necessarily designed by systematically considering the data structure, similarity metric, discriminant subspace, and classifier. In this paper, we provide evidence that, besides the Fisher criterion, graph embedding, and tensorization used in many existing methods, the correlation-based similarity metric embodied in supervised multilinear discriminant subspace learning can additionally improve the classification performance. In particular, a novel discriminant subspace learning algorithm, called correlation tensor analysis (CTA), is designed to incorporate both graph-embedded correlational mapping and discriminant analysis in a Fisher type of learning manner. The correlation metric can estimate intrinsic angles and distances for the locally isometric embedding, which can deal with the case when Euclidean metric is incapable of capturing the intrinsic similarities between data points. CTA learns multiple interrelated subspaces to obtain a low-dimensional data representation reflecting both class label information and intrinsic geometric structure of the data distribution. Extensive comparisons with most popular subspace learning methods on face recognition evaluation demonstrate the effectiveness and superiority of CTA. Parameter analysis also reveals its robustness.

Estimating human age automatically via facial image analysis has lots of potential real-world applications, such as human computer interaction and multimedia communication. However, it is still a challenging problem for the existing computer vision systems to automatically and effectively estimate human ages. The aging process is determined by not only the person's gene, but also many external factors, such as health, living style, living location, and weather conditions. Males and females may also age differently. The current age estimation performance is still not good enough for practical use and more effort has to be put into this research direction. In this paper, we introduce the age manifold learning scheme for extracting face aging features and design a locally adjusted robust regressor for learning and prediction of human ages. The novel approach improves the age estimation accuracy significantly over all previous methods. The merit of the proposed approaches for image-based age estimation is shown by extensive experiments on a large internal age database and the public available FG-NET database.

Recently, extensive studies on human faces in the human-computer interaction (HCI) field reveal significant potentials for designing automatic age estimation systems via face image analysis. The success of such research may bring in many innovative HCI tools used for the applications of human-centered multimedia communication. Due to the temporal property of age progression, face images with aging features may display some sequential patterns with low-dimensional distributions. In this paper, we demonstrate that such aging patterns can be effectively extracted from a discriminant subspace learning algorithm and visualized as distinct manifold structures. Through the manifold method of analysis on face images, the dimensionality redundancy of the original image space can be significantly reduced with subspace learning. A multiple linear regression procedure, especially with a quadratic model function, can be facilitated by the low dimensionality to represent the manifold space embodying the discriminative property. Such a processing has been evaluated by extensive simulations and compared with the state-of-the-art methods. Experimental results on a large size aging database demonstrate the effectiveness and robustness of our proposed framework.

Emotive audio-visual avatars are virtual computer agents which have the potential of improving the quality of human-machine interaction and human-human communication significantly. However, the understanding of human communication has not yet advanced to the point where it is possible to make realistic avatars that demonstrate interactions with natural-sounding emotive speech and realistic-looking emotional facial expressions. In this paper, We propose the various technical approaches of a novel multimodal framework leading to a text-driven emotive audio-visual avatar. Our primary work is focused on emotive speech synthesis, realistic emotional facial expression animation, and the co-articulation between speech gestures (i.e., lip movements) and facial expressions. A general framework of emotive text-to-speech (TTS) synthesis using a diphone synthesizer is designed and integrated into a generic 3-D avatar face model. Under the guidance of this framework, we therefore developed a realistic 3-D avatar prototype. A rule-based emotive TTS synthesis system module based on the Festival-MBROLA architecture has been designed to demonstrate the effectiveness of the framework design. Subjective listening experiments were carried out to evaluate the expressiveness of the synthetic talking avatar.

Advances in computer processing power and emerging algorithms are allowing new ways of envisioning human-computer interaction. Although the benefit of audio-visual fusion is expected for affect recognition from the psychological and engineering perspectives, most of existing approaches to automatic human affect analysis are unimodal: information processed by computer system is limited to either face images or the speech signals. This paper focuses on the development of a computing algorithm that uses both audio and visual sensors to detect and track a user's affective state to aid computer decision making. Using our multistream fused hidden Markov model (MFHMM), we analyzed coupled audio and visual streams to detect four cognitive states (interest, boredom, frustration and puzzlement) and seven prototypical emotions (neural, happiness, sadness, anger, disgust, fear and surprise). The MFHMM allows the building of an optimal connection among multiple streams according to the maximum entropy principle and the maximum mutual information criterion. Person-independent experimental results from 20 subjects in 660 sequences show that the MFHMM approach outperforms face-only HMM, pitch-only HMM, energy-only HMM, and independent HMM fusion, under clean and varying audio channel noise condition.

Factor analysis (FA) seeks to reveal the relationship between an observed vector variable and a latent variable of reduced dimension. It has been widely used in many applications involving high-dimensional data, such as image representation and face recognition. An intrinsic limitation of FA lies in its potentially poor performance when the data dimension is high, a problem known as curse of dimensionality. Motivated by the fact that images are inherently matrices, we develop, in this brief, an FA model for matrix-variate variables and present an efficient parameter estimation algorithm. Experiments on both toy and real-world image data demonstrate that the proposed matrix-variant FA model is more efficient and accurate than the classical FA approach, especially when the observed variable is high-dimensional and the samples available are limited.

We propose a complete solution to full modality person-profiling for speakers and submodality person-profiling for non-speakers in real-world videos. This is a step towards building an elaborate database efface, name and voice correspondence for speakers appearing in the news videos. In addition we are also interested in only name and face correspondence database for non-speakers who appear during voice-overs. We use an unsupervised technique for creating a speaker identification database and a unique primary feature matching and parallel line matching algorithm for creating a non-speaker identification database. We tested our approach on real world data and the results show good performance for news videos. It can be incorporated as part of a larger multimedia news video analysis system or a multimedia search system for efficient news video retrieval and browsing.

Subspace learning based face recognition methods have attracted considerable interests in recently years, including principal component analysis (PCA), linear discriminant analysis (LDA), locality preserving projection (LPP), neighborhood preserving embedding (NPE), marginal fisher analysis (MFA) and local discriminant embedding (LDE). These methods consider an n1timesn2image as a vector in Rn1timesn2and the pixels of each image are considered as independent. While an image represented in the plane is intrinsically a matrix. The pixels spatially close to each other may be correlated. Even though we have n1xn2pixels per image, this spatial correlation suggests the real number of freedom is far less. In this paper, we introduce a regularized subspace learning model using a Laplacian penalty to constrain the coefficients to be spatially smooth. All these existing subspace learning algorithms can fit into this model and produce a spatially smooth subspace which is better for image representation than their original version. Recognition, clustering and retrieval can be then performed in the image subspace. Experimental results on face recognition demonstrate the effectiveness of our method.

The fusion of audio and visual speech is an instance of the general sensory fusion problem. The sensory fusion problem arises in the situation when multiple channels carry complementary information about different components of a system. In the case of audio-visual speech, the two modalities manifest two aspects of the same underlying speech production process. From an observer's view, the audio channel and the visual channel represent two interacting stochastic processes. We seek a framework that can model the two individual processes as well as their dynamic interactions. One interesting aspect of audio-visual speech is the inherent asynchrony between the audio and visual channels. Most early integration approaches to the fusion problem assume tight synchrony between the two. However, studies have shown that human perception of bimodal speech does not require rigid synchronization of the two modalities. Furthermore, humans appear to use the audio-visual asynchronies as multimodal features. For example, it is well known that the voice onset time is an important cue to the voicing feature in stop consonants. This information can be conveyed bimodally by the interval between seeing the stop release and hearing the vocal cord vibration. Therefore, a successful fusion scheme should not only be tolerant to asynchrony between the audio and visual cues, but also be apt to capture and exploit this bimodal feature.

We present the Conformal Embedding Analysis (CEA) for feature extraction and dimensionality reduction. Incorporating both conformal mapping and discriminating analysis, CEA projects the high-dimensional data onto the unit hypersphere and preserves intrinsic neighbor relations with local graph modeling. Through the embedding, resulting data pairs from the same class keep the original angle and distance information on the hypersphere, whereas neighboring points of different class are kept apart to boost discriminating power. The subspace learned by CEA is gray-level variation tolerable since the cosine-angle metric and the normalization processing enhance the robustness of the conformal feature extraction. We demonstrate the effectiveness of the proposed method with comprehensive comparisons on visual classification experiments.

Interactively learning from a small sample of unlabeled examples is an enormously challenging task, one that often arises in vision applications. Relevance feedback and more recently active learning are two standard techniques that have received much attention towards solving this interactive learning problem. How to best utilize the user's effort for labeling, however, remains unanswered. It has been shown in the past that labeling a diverse set of points is helpful, however, the notion of diversity has either been dependent on the learner used, or computationally expensive. In this paper, we intend to address these issues in the bipartite ranking setting. First, we introduce a scheme for picking the query set which will be labeled by an oracle so that it will aid us in learning the ranker in as few active learning rounds as possible. Secondly, we propose a fundamentally motivated, information theoretic view of diversity and its use in a fast, non-degenerate active learning-based relevance feedback setting. Finally, we report comparative testing and results in a real-time image retrieval setting.

A large family of algorithms for dimensionality reduction end with solving a Trace Ratio problem in the form of arg maxWTr(WTSPW)/Tr(WT SIW)1, which is generally transformed into the corresponding Ratio Trace form arg maxWTr[ (WTSIW)-1(WTSPW) ] for obtaining a closed-form but inexact solution. In this work, an efficient iterative procedure is presented to directly solve the Trace Ratio problem. In each step, a Trace Difference problem arg maxWTr [WT(SP- lambdaSI) W] is solved with lambda being the trace ratio value computed from the previous step. Convergence of the projection matrix W, as well as the global optimum of the trace ratio value lambda, are proven based on point-to-set map theories. In addition, this procedure is further extended for solving trace ratio problems with more general constraint WTCW=I and providing exact solutions for kernel-based subspace learning problems. Extensive experiments on faces and UCI data demonstrate the high convergence speed of the proposed solution, as well as its superiority in classification capability over corresponding solutions to the ratio trace problem.

The success of tensor-based subspace learning depends heavily on reducing correlations along the column vectors of the mode-k flattened matrix. In this work, we study the problem of rearranging elements within a tensor in order to maximize these correlations, so that information redundancy in tensor data can be more extensively removed by existing tensor-based dimensionality reduction algorithms. An efficient iterative algorithm is proposed to tackle this essentially integer optimization problem. In each step, the tensor structure is refined with a spatially-constrained Earth Mover's Distance procedure that incrementally rearranges tensors to become more similar to their low rank approximations, which have high correlation among features along certain tensor dimensions. Monotonic convergence of the algorithm is proven using an auxiliary function analogous to that used for proving convergence of the Expectation-Maximization algorithm. In addition, we present an extension of the algorithm for conducting supervised subspace learning with tensor data. Experiments in both unsupervised and supervised subspace learning demonstrate the effectiveness of our proposed algorithms in improving data compression performance and classification accuracy.

Multimodal sensing, data processing, and system integration are still in a fast growing stage of research and development. Much work needs to be done to ensure successful real-world, large-scale surveillance applications. The collection in this book is just samples of some representative works in the field and therefore by no means to be complete. However, we hope this collection will stimulate more interests in the research and development of multimodal surveillance techniques, including the further developments of multimodal sensors, multimodal data fusion algorithms, and multimodal surveillance systems.

In most machine learning and pattern recognition problems, the large number of high-dimensional sensory data, such as images and videos, are often labeled manually for training classifiers and modeling features, which is time-consuming and tedious. To automatically execute this process by machine, we present the unsupervised high-dimensional data clustering and automatic labeling algorithms, called locally embedded clustering (LEC): (i) constructing the neighborhood weighted graph with an appropriate distance metric; (ii) tuning the regularization parameter to smooth the approximated manifold; (iii) calculating the unified projection in a closed-form solution for the embedding and dimensionality reduction; (iv) choosing the top or bottom coordinates of the embedded low-dimensional space for data representation; (v) normalizing the low-dimensional representation to have unit length; (vi) clustering and labeling the data via K-means. Experimental results demonstrate that LEC provides better data representation, more efficient dimensionality reduction and better clustering performance than many existing methods.

How to encode a face is a widely studied problem in both pattern recognition and psychology literatures. Many feature descriptors, Gabor feature, local binary pattern (LBP), and edge orientation histogram, have been proposed. In this paper, we give a comprehensive study of these descriptors under the framework of principal component analysis (PCA) followed by linear discriminant analysis (LDA), compared on three different popular similarity measures and two different feature correspondence strategies: holistic and local. Moreover, we present a new feature descriptor named multi-radius LBP, and also propose a combination scheme for the LBP and Gabor descriptor. The experiments on the Purdue and CMU PIE databases demonstrate that 1) an obvious recognition boost of LBP is achieved under PCA+LDA framework compared to the direct NN classification; 2) the LBP and Gabor features are comparable as well as mutually complementary, and the combination of these two descriptors brings a significant improvement in classification capability over single ones; and 3) the multi-radius LBP shows to outperform all the state-of-the-art feature descriptors.

Building face models is an essential task in face recognition, tracking and etc. However, most of the current techniques require hand-labelling or special machinery such as cyber-scanner to extract the face model. In the paper, we propose an unsupervised algorithm to learn the face texture from video. The proposed approach models the video sequence as a mixture of dynamic face-layers and background layers, where the dynamic face-layers may undergo 3D motions in the video. The hidden variables and their generating process is represented by probabilistic graphical model. The model is learnt by EM algorithm with variational approximation. The proposed approach offers several advantage over existing algorithms. First, it derive its learning power by a generative model which naturally represents the generating process of videos. Second, it does not require any labelling or face detection algorithm. Third, the application domain of the proposed algorithm is not restricted to extracting face texture and it can be adapted to model other objects as well. The experimental results demonstrate that our model is capable of learning the appearance model of faces with complex 3D motions in the video.

Local optimum has been one of the most difficult problems in image registration notwithstanding the extensive research effort that has been put into solving it. Local optimums occur when a portion of patterns in the floating image coincide with a portion of patterns in the reference image even though the two images are not entirely matched. Existing hierarchical or multi-scale methods suffer from this problem mainly because some redundant information that causes local optimums appears in multiple scales. We propose to avoid it by decomposing an image into several mutually exclusive scale components so that minimal redundant information is present. Our method is evaluated and compared with existing methods using high resolution satellite imagery where thousands of local optimum traps are hidden. We show that our method has significant improvement over existing solutions in both robustness and efficiency.

In this paper, we take the human age and pose estimation problems as examples to study automatic designing regressor from training samples with uncertain nonnegative labels. First, the nonnegative label is predicted as the square norm of a matrix, which is bilinearly transformed from the nonlinear mappings of the candidate kernels. Two transformation matrices are then learned for deriving such a matrix by solving a semi definite programming (SDP) problem, in which the uncertain label of each sample is expressed as two inequality constraints. The objective function of SDP controls the ranks of these two matrices, and consequently automatically determines the structure of the regressor. The whole framework for automatic designing regressor from samples with uncertain nonnegative labels has the following characteristics: 1) SDP formulation makes full use of the uncertain labels, instead of using conventional fixed labels; 2) regression with matrix norm naturally guarantees the nonnegativity of the labels, and greater prediction capability is achieved by integrating the squares of the matrix elements, which act as weak regressors; and 3) the regressor structure is automatically determined by the pursuit of simplicity, which potentially promotes the algorithmic generalization capability. Extensive experiments on two human age databases, FG-NET and Yamaha, as well as the Pointing'04 pose database, demonstrate encouraging estimation accuracy improvements over conventional regression algorithms.

We solve the semi-supervised multi-class object classification problem by a graph-based learning algorithm, called Laplacian affinity propagation (LAP). The idea is to model and train both labeled and unlabeled data by constructing a local neighborhood affinity graph in a smoothness formulation of Laplacian matrix, based on graph mincuts or harmonic energy minimization. The unknown labels for unlabeled data are inferred from an optimized graph embedding procedure subject to the labeled data. Such label-to-unlabel propagation scheme can provide a closed form solution via a learning framework that is flexible for any new design. LAP integrates embedding and classifier together and gives smooth labels with respect to the underlying manifold structure formed by the training data. Object classification experiments on COIL database demonstrate the effectiveness and applicability of such algorithm.

We present a novel expert-model of Query-Driven Locally Adaptive (QDLA) Fisher faces for robust face recognition. For each query face, the proposed method first fits local Fisher models with different appearances. A hybrid expert model then integrates these local models and combines the classification results based on the estimated error rate for each local model. This approach addresses the large size recognition problem, where many local variations can not be adequately handled by a single global model in a single appearance space. To speed up the query process, Locality Sensitive Hash (LSH) is applied for fast nearest neighbor search. Experiments demonstrate the approach to be effective, robust, and fast for large size, multi-class, and multi-variance data sets.

The major problem in building a good lipreading system is to extract effective visual features from the enormous quantity of video sequences data. For appearance-based feature analysis in lipreading, classical methods, e.g. DCT, PCA and LDA, are usually applied to dimensionality reduction. We present a new pattern classification algorithm, called locality discriminant graph (LDG), and develop a novel lipreading framework to successfully apply LDG to the problem. LDG takes the advantages of both manifold learning and Fisher criteria to seek the linear embedding which preserves the local neighborhood affinity withinsameclasswhile discriminating the neighborhood amongdifferentclasses. The LDG embedding is computed in closed-form and tuned by the only open parameter of k-NN number. Experiments on AVICAR corpus provide evidence that the graph-based pattern classification methods can outperform classical ones for lipreading.

This paper proposes an approach to searching human behaviors in videos using spatial-temporal words which are learnt from unlabelled data with various human behaviors through unsupervised learning. Both the query and the searched videos are represented by codewords frequencies, which capture the intrinsic information of motion and appearance of human behaviors. This representation further enables us to make use of integral histograms to accelerate the searching procedure. The performance also benefits from our feature representation that, through a MAX-like operation, may simulate the cortical equivalent of the machine-vision "window of analysis"(M. Riesenhuber and T. Poggio, 1999). Examples of challenging sequences with complex behaviors, including tennis and ballet, are shown.

This paper introduces a head pose estimation system that localizes nose-tip of the faces and estimate head poses in images simultaneously. After the nose-tip in the training data are manually labeled, the appearance variation caused by head pose changes is characterized by tensor model. Given images with unknown head pose and nose-tip location, the nose-tip of the face is localized in a coarse-to-fine fashion, and the head pose can be estimated simultaneously. We evaluated our system on the Pointing'04 head pose image database with 50% of the data as training set and the rest as testing set. With the nose-tip location provided, our head pose estimators can achieve 94% head pose classification accuracy(within plusmn15deg). With nose-tip automatically localized, we achieves 85% nose-tip localization accuracy(within 3 pixels from the ground truth), and 81% head pose classification accuracy (within plusmn15deg).

Automatically discovering common visual patterns in images is very challenging due to the uncertainties in the visual appearances of such spatial patterns and the enormous computational cost involved in exploring the huge solution space. Instead of performing exhaustive search on all possible candidates of such spatial patterns at various locations and scales, this paper presents a novel and very efficient algorithm for discovering common visual patterns by designing a provably correct and computationally efficient pruning procedure that has a quadratic complexity. This new approach is able to efficiently search a set of images for unknown visual patterns that exhibit large appearance variations because of rotation, scale changes, slight view changes, color variations and partial occlusions.

The need to analyze and index large amounts of video information is becoming more important as the way people consume media continues to change. In recent years, the push to attack multimedia indexing and retrieval applications in a holistic, multi-modal way has garnered great attention. In this work we propose the holistic use of both audio, visual and textual information for the automatic indexing of broadcast news video to create person-profiles. Indexing videos in this matter helps facilitate a unique way to create multimedia person databases automatically, as well as for attacking existing video analysis tasks. We test our algorithm on news data from NBC and present areas for future exploration.

A novel framework of multimodal human-machine or human-human interaction via real-time humanoid avatar communication is proposed for real-world mobile application. It integrates audio-visual analysis and synthesis modules to realize real-time head tracking, multichannel and runtime animations, visual TTS and real-time viseme detection and rendering. The 3D avatar provides customized modeling for low-bit rate virtual communication by adopting M3G standard and supports MPEG-4 FAPs. A robust user head tracker and the associated head pose and motion estimation scheme are developed for real-time avatar animation control at remote locations. The framework is recognized as an effective design for realistic industrial products of human-to-human mobile communication.

Extensive recent studies on human faces reveal significant potential applications of automatic age estimation via face image analysis. Due to the temporal features of age progression, aging face images display sequential pattern of low-dimensional distribution. Through manifold analysis of face pictures, we developed a novel age estimation framework. The manifold learning methods are applied to find a sufficient embedding space and model the low-dimensional manifold data with a multiple linear regression function. Experimental results on a large size age database demonstrate the effectiveness of the framework. To our best knowledge, this is the first work involving the manifold ways of age estimation.

Head pose appearances under the pan and tilt variations span a high dimensional manifold that has complex structures and local variations. For pose estimation purpose, we need to discover the subspace structure of the manifold and learn discriminative subspaces/metrics for head pose recognition. The performance of the head pose estimation is heavily dependent on the accuracy of structure learnt and the discriminating power of the metric. In this work we develop a query point driven, localized linear subspace learning method that approximates the non-linearity of the head pose manifold structure with piece-wise linear discriminating subspaces/metrics. Simulation results demonstrate the effectiveness of the proposed solution in both accuracy and computational efficiency.

Speaker verification is a technology of verifying the claimed identity of a speaker based on the speech signal from the speaker (voice print). To learn the score of similarity between each pair of target and trial utterances, we investigated two different discriminative learning frameworks: Fisher mapping followed by SVM learning and utterance transform followed by iterative cohort modeling (ICM). In both methods, a mapping is applied to map speech utterance from a variable-length acoustic feature sequence into a fixed dimensional vector. SVM learning constructs a classifier in the mapped vector space for speaker verification. ICM learns a metric in this vector space by incorporating discriminative learning methods. The obtained metric is then used by a nearest neighbor classifier for speaker verification. The experiments conducted on NIST02 corpus show that both discriminative learning methods outperform the baseline GMM-UBM system. Furthermore, we observe that the ICM-based method is more effective than the SVM-based method, indicating that the metric learning scheme is more powerful in constructing a better metric in the mapped vector space.

Most techniques for image analysis consider the image labels fixed and without uncertainty. In this paper, we address the problem of ordinal/rank label prediction based on training samples with uncertain labels. First, the core ranking model is designed as the bilinear fusing of multiple candidate kernels. Then, the parameters for feature selection and kernel selection are learned by maximum a posteriori for given samples and uncertain labels. The convergency provable Expectation-Maximization (EM) method is used for inferring these parameters. The effectiveness of the proposed algorithm is finally validated by the extensive experiments on age ranking task. The FG-NET and Yamaha aging database are used for the experiments, and our algorithm significantly outperforms those state-of-the-art algorithms ever reported in literature.

Mismatch between training and testing data is a major error source for both automatic speech recognition (ASR) and automatic speaker identification (ASI). In this paper, we first present a statistical weighting concept to exploit the unequal sensitivity of mel-frequency cepstral coefficients (MFCC) components to against the mismatch, such as ambient noise, recording equipment, transmission channels, and inter-speaker variations. We further design a new Kullback-Leibler (KL) distance based weighting algorithm according to the proposed weighting concept to real-world problems in which the label information is often not provided. We examine our algorithm in ASR with mismatch by different speakers and also in ASI with mismatch by channel noises. Experimental results demonstrate the effectiveness and robustness of our proposed method.

We propose a discriminative learning approach for fusing multichannel sequential data with application to detect unsafe driving patterns from multi-channel driving recording data. The fusion is performed using a discriminatively trained graphical model -conditional random field (CRF). The proposed approach offers several advantage over existing information fusing approaches. First, it derives its classification power by directly modelling and maximizing the conditional probability. Second, it represents the variable dependency in an undirected graph, which is very efficient in inference. Third, it does not require to label all the training data and utilizes both labelled and unlabelled data efficiently by semi-supervised learning algorithms. The proposed approach is evaluated on driving recording data collected from driving simulator -STISIM. Experiments show it outperforms the simple discriminative classifier (SVM) and generative model (HMM).

Trajectories of moving objects provide crucial clues for video event analysis especially in surveillance applications. In this paper, we study the problem of detecting anomalous events by analyzing the motion trajectories in videos. Different trajectories of the same category may have varying relative velocities, in addition to the variations and noises in location samples; hence the core of the problem is to provide a robust and accurate function for measuring the similarities of trajectory pairs. We propose a novel learning based algorithm for estimating the similarities of the multi-dimensional sequence pairs, and then an anomaly detection framework is presented to detect anomalous motion trajectories in surveillance videos. Our proposed algorithm offers several advantages over the traditional algorithms for dealing with the trajectories of moving objects. First, the similarity measurement is robust against data imperfections such as noise, algorithmic error and etc. Second, we introduce a learning algorithm which allows the similarity function to be adapted to the particular problems being solved. Third, the proposed anomaly detection framework is fully automatic and without parametric distribution assumption on the data. The experiments on both outdoor and indoor surveillance videos validate the effectiveness of our proposed framework in detecting anomalous trajectories.

Automated analysis of human affective behavior has attracted increasing attention from researchers in psychology, computer science, linguistics, neuroscience, and related disciplines. Promising approaches have been reported, including automatic methods for facial and vocal affect recognition. However, the existing methods typically handle only deliberately displayed and exaggerated expressions of prototypical emotions-despite the fact that deliberate behavior differs in visual and audio expressions from spontaneously occurring behavior. Recently efforts to develop algorithms that can process naturally occurring human affective behavior have emerged. This paper surveys these efforts. We first discuss human emotion perception from a psychological perspective. Next, we examine the available approaches to solving the problem of machine understanding of human affective behavior occurring in real-world settings. We finally outline some scientific and engineering challenges for advancing human affect sensing technology.

In this paper, we present a novel semisupervised regression algorithm working on multiclass data that may lie on multiple manifolds. Unlike conventional manifold regression algorithms that do not consider the class distinction of samples, our method introduces the class information to the regression process and tries to exploit the similar configurations shared by the label distribution of multi-class data. To utilize the correlations among data from different classes, we develop a cross-manifold label propagation process and employ labels from different classes to enhance the regression performance. The interclass relations are coded by a set of intermanifold graphs and a regularization item is introduced to impose inter-class smoothness on the possible solutions. In addition, the algorithm is further extended with the kernel trick for predicting labels of the out-of-sample data even without class information. Experiments on both synthesized data and real world problems validate the effectiveness of the proposed framework for semisupervised regression.

Due to physiology and linguistic difference between speakers, the spectrum pattern for the same phoneme of two speakers can be quite dissimilar. Without appropriate alignment on the frequency axis, the misalignment will reduce the modeling efficiency resutling in performance degradation. In this paper, a novel data-driven framework is proposed to build the alignment of the frequency axes of two speakers. This alignment between two frequency axes is essentially a frequency domain correspondence of the two speakers. To establish the correspondence, we formulate the task as a global optimal matching problem. The local matching of frequency bins is achieved by comparing the local feature of the spectrogram along the frequency bins. The local feature is actually capturing the local pattern in the spectrogram. Given the local matching score, a dynamic programming is then applied to find the optimal correspondence. Experiments on TIMIT corpus and TIDIGITS corpus clearly show the effectiveness of this method. 1.

In recent years, spectral clustering method has gained attentions because of its superior performance compared to other traditional clustering algorithms such as K-means algorithm. The existing spectral clustering algorithms are all off-line algorithms, i.e., they can not incrementally update the clustering result given a small change of the data set. However, the capability of incrementally updating is essential to some applications such as real time monitoring of the evolving communities of websphere or blogsphere. Unlike traditional stream data, these applications require incremental algorithms to handle not only insertion/deletion of data points but also similarity changes between existing items. This paper extends the standard spectral clustering to such evolving data by introducing the incidence vector/matrix to represent two kinds of dynamics in the same framework and by incrementally updating the eigenvalue system. Our incremental algorithm, initialized by a standard spectral clustering, continuously and efficiently updates the eigenvalue system and generates instant cluster labels, as the data set is evolving. The algorithm is applied to a blog data set. Compared with recomputation of the solution by standard spectral clustering, it achieves similar accuracy but with much lower computational cost. Close inspection into the blog content shows that the incremental approach can discover not only the stable blog communities but also the evolution of the individual multi-topic blogs.

A novel head tracking driven camera mouse system, called "hMouse", is developed for manipulating hand-free perceptual user interfaces. The system consists of a robust real-time head tracker, a head pose/motion estimator, and a virtual mouse control module. For the hMouse tracker, we propose a 2D detection/tracking complementary switching strategy with an interactive loop. Based on the reliable tracking results, hMouse calculates the user's head roll, tilt, yaw, scaling, horizontal, and vertical motion for further mouse control. Cursor position is navigated and fine tuned by calculating the relative position of tracking window in image space and the user's head tilt or yaw rotation. After mouse cursor is navigated to the desired location, head roll rotation triggers virtual mouse button clicks. Experimental results demonstrate that hMouse succeeds under the circumstances of user jumping, extreme movement, large degree rotation, turning around, hand/object occlusion, part face out of camera shooting region, and multiuser occlusion. It provides alternative solutions for convenient device control, which encourages the application of interactive computer games, machine guidance, robot control, and machine access for disabilities and elders

In order to avoid the notorious drifting problem for tracking system, a new integrated appearance learning framework is proposed in this paper. Previous tracking frameworks with appearance learning ability (Black, et al., 1998) either require supervised offline training or will fail inevitably if the tracker locks on the background. While in our framework, no offline training is required. Given the location of the object in the first frame of the video sequence, we model the foreground (the image patch containing the object)/background difference as the transition cost in our tracking objective function. A tracker based on dynamic programming (DP) and template prediction (Toyama, et al., 2001) is carried out on the pixels with high foreground-likelihood. The typical views (i.e. appearance model) proposed by the tracker are used to initialize the states of a hidden Markov model (HMM). With the learned HMM, the tracking results and the appearance model can be further refined until the video sequence and all of these estimated parameters/hidden variables can be well explained by the HMM. Through this iterative procedure, typical views of the object, transition probabilities between the typical views, and location of the object are simultaneously estimated with strong confidence. The experiments show that the proposed framework achieves fairly satisfied results for several challenging video sequences and therefore has many potential applications for video analysis

In this paper, a novel hybrid dimension reduction technique for classification is proposed based on the hybrid analysis of principal component analysis (PCA) and linear discriminant analysis (LDA). LDA is known for capturing the most discriminant features of the data in the projected space while PCA is known for preserving the most descriptive ones after projection. Our hybrid technique integrates discriminant and descriptive information and finds a richer set of alternatives beyond LDA and PCA in a 2-D parametric space, which fits a specific classification task and data distribution better. Theoretical study shows that our technique also alleviates the singularity problem of scatter matrix, which is caused by small training set, and increases the effective dimension of the projected subspace. In order to find the hybrid features adaptively and avoid exhaustive parameter searching, we further propose a boosted hybrid analysis method that incorporates a nonlinear boosting process to enhance a set of hybrid classifiers and combine them into a more accurate one. Compared with the other techniques that aim at combining PCA and LDA, our approaches are novel because our method finds alternatives to LDA and PCA in a 2-D parameter space and the boosting process provides enhancement and robust combination of the classifiers. Extensive experiments are conducted on benchmark and real image databases to compare our proposed methods with the state-of-the-art linear and nonlinear discriminant analysis techniques. The results show the superior performance of our hybrid analysis methods

Supervised subspace learning techniques have been extensively studied in biometrics literature; however, there is little work dedicated to: 1) how to automatically determine the subspace dimension in the context of supervised learning, and 2) how to explicitly guarantee the classification performance on a training set. In this paper, by following our previous work on unified subspace learning framework in our earlier work, we present a general framework, called parameter-free graph embedding (PFGE) to solve the above two problems by posing a general supervised subspace learning task as a semidefinite programming problem. The semipositive feature Gram matrix, namely the product of the transformation matrix and its transpose, is derived by optimizing a trace difference form of an objective function extended from that in our earlier work with the constraints that guarantee the class homogeneity within the neighborhood of each datum. Then, the subspace dimension and the feature weights are simultaneously obtained via the singular value decomposition of the feature Gram matrix. In addition, to alleviate the computational complexity, the Kronecker product approximation of the feature Gram matrix is proposed by taking advantage of the essential matrix form of image pixels. The experiments on simulated data and real-world data demonstrate the capability of the new PFGE framework in estimating the subspace dimension for supervised learning as well as the superiority in classification performance over traditional algorithms for subspace learning

This paper presents a unified solution to three unsolved problems existing in face verification with subspace learning techniques: selection of verification threshold, automatic determination of subspace dimension, and deducing feature fusing weights. In contrast to previous algorithms which search for the projection matrix directly, our new algorithm investigates a similarity metric matrix (SMM). With a certain verification threshold, this matrix is learned by a semidefinite programming approach, along with the constraints of the kindred pairs with similarity larger than the threshold, and inhomogeneous pairs with similarity smaller than the threshold. Then, the subspace dimension and the feature fusing weights are simultaneously inferred from the singular value decomposition of the derived SMM. In addition, the weighted and tensor extensions are proposed to further improve the algorithmic effectiveness and efficiency, respectively. Essentially, the verification is conducted within an affine subspace in this new algorithm and is, hence, called the affine subspace for verification (ASV). Extensive experiments show that the ASV can achieve encouraging face verification accuracy in comparison to other subspace algorithms, even without the need to explore any parameters.

The paper proposes an adaptive retrieval approach based on the concept of relevance-feedback, which establishes a link between high-level concepts and low-level features, using the user's feedback not only to assign proper weights to the features, but also to dynamically select them within a large collection of parameters. The target is to identify a set of relevant features according to a user query while at the same time maintaining a small sized feature vector to attain better matching and lower complexity. To this end, the image description is modified during each retrieval by removing the least significant features and better specifying the most significant ones. The feature adaptation is based on a hierarchical approach. The weights are then adjusted based on previously retrieved relevant and irrelevant images without further user-feedback. The algorithm is not fixed to a given feature set. It can be used with different hierarchical feature sets, provided that the hierarchical structure is defined a priori. Results achieved on different image databases and two completely different feature sets show that the proposed algorithm outperforms previously proposed methods. Further, it is experimentally demonstrated that it approaches the results obtained by state-of-the-art feature-selection techniques having complete knowledge of the data set.

The ability of a computer to detect and appropriately respond to changes in a user's affective state has significant implications to human-computer interaction (HCI). In this paper, we present our efforts toward audio-visual affect recognition on 11 affective states customized for HCI application (four cognitive/motivational and seven basic affective states) of 20 nonactor subjects. A smoothing method is proposed to reduce the detrimental influence of speech on facial expression recognition. The feature selection analysis shows that subjects are prone to use brow movement in face, pitch and energy in prosody to express their affects while speaking. For person-dependent recognition, we apply the voting method to combine the frame-based classification results from both audio and visual channels. The result shows 7.5% improvement over the best unimodal performance. For person-independent test, we apply multistream HMM to combine the information from multiple component streams. This test shows 6.1% improvement over the best component performance

We present a novel framework to automatically separate brain region from other non-brain regions in head images. The idea of the proposed method is to estimate larger scale patterns in an image and then correct the boundaries iteratively. The scale estimation is based on the recently proposed total variation (TV) regularized L^1 functional. An iterative optimization method is used to refine non-convex and acute angle boundaries. The final algorithm is able to extract large scale patterns with arbitrary shapes, which is particularly suitable for brain extraction. In order to reduce the computation overhead in 3D data, a multi-level technique is proposed to exponentially improve the speed of the brain extraction process. Based on accurate results of brain extraction, a non-rigid brain registration algorithm is proposed to improve accuracy and consistency of existing registration methods. Experimental results on real 3D brain MR images demonstrate that the proposed methods outperform existing solutions. In addition, results are provided to show that the proposed algorithm can also be used to segment large scale patterns in general images.

An efficient Nonparametric Belief Propagation (NBP) algorithm is developed in this paper. While the recently proposed nonparametric belief propagation algorithm has wide applications such as articulated tracking [22, 19], superresolution [6], stereo vision and sensor calibration [10], the hardcore of the algorithm requires repeatedly sampling from products of mixture of Gaussians, which makes the algorithm computationally very expensive. To avoid the slow sampling process, we applied mixture Gaussian density approximation by mode propagation and kernel fitting [2, 7]. The products of mixture of Gaussians are approximated accurately by just a few mode propagation and kernel fitting steps, while the sampling method (e.g. Gibbs sampler) needs many samples to achieve similar approximation results. The proposed algorithm is then applied to articulated body tracking for several scenarios. The experimental results show the robustness and the efficiency of the proposed algorithm. The proposed efficient NBP algorithm also has potentials in other applications mentioned above.

Forging the special documents such as driver's license, visa and passport, has reached unprecedented level in recent years. In order to stop such crimes, various methods have been presented. One typical way is to do human identification if face picture is present in the documents. But it’s often found that criminals modify the face picture in the documents, thus card validation purely based on face picture may also not be reliable. In this paper we present a novel method for such document authentication problem, we propose to do face image compression and store the data as barcode in an identification document, both of the barcode and original photo are printed in the document. By comparing the face image scanned from the document and the picture reconstructed from barcode, using different image processing methods to enhance the image and Gabor wavelet features, we can make a decision whether the document is valid or not. Index Terms -- Image enhancement, Identification, Data compression, Matching.

This paper studies the speech of three talkers with spastic dysarthria caused by cerebral palsy. All three subjects share the symptom of low intelligibility, but causes differ. First, all subjects tend to reduce or delete word-initial consonants; one subject deletes all consonants. Second, one subject exhibits a painstaking stutter. Two algorithms were used to develop automatic isolated digit recognition systems for these subjects. HMM-based recognition was successful for two subjects, but failed for the subject who deletes all consonants. Conversely, digit recognition experiments assuming a fixed word length (using SVMs) were successful for two subjects, but failed for the subject with the stutter.

In this paper we address the problem of document retrieval with semantically structured queries - queries where each term has a tagged field label. We introduce Dirichlet Aspect Weighting model which integrates terms from external databases into the query language model in a bayesian learning framework. For this model, the Dirichlet prior distribution is governed by parameters which depend on the number of fields in the external databases. This model needs additional examples to be augmented to the semantically structured query. These examples are obtained using pseudo relevance feedback. We formulate a loglikelihood function for the Dirichlet Aspect Weighting model and maximize it using a novel Generalized EM algorithm. Comparison of the results of Dirichlet Aspect Weighting model on TREC 2005 Genomics Track dataset with baseline methods using pseudo relevance feedback, while incorporating terms from external databases shows an improvement.

In this paper, we study the calibration of head pose estimation in stereo camera setting for meeting room video event analysis. Head pose information infers the direction of attention of the subjects in video, therefore is valuable for video event analysis/indexing, especially in meeting room scenario. We are developing a multi-modal meeting room data analyzing system for studying meeting room interaction dynamics, in which head pose estimation is one of the key components. As each subject in the meeting room can be observed by a pair of stereo cameras, we do 2D head tracking for the subject in each camera, and the 3D coordinate of the head can be obtained by triangulation. The 3D head pose is estimated in one of the camera coordinate system, we develop a procedure to accurately convert the estimated 3D pose in the camera coordinate system to that in the world coordinate system. In the experiment, visualization of the estimated head pose and location in world coordinate system verifies the soundness of our design. The estimated head pose and 3D location of the subjects in the meeting room allows further analysis of meeting room interaction dynamics, such as F-formation, floor-control, etc.

Multi-modal person authentication systems can achieve higher performance and robustness by combining different modalities. The current fusion strategies of different modalities are mainly based on the output of individual modalities. However, there are detail structures between facial movement and speech signal. In this paper, audio/visual association, a lower level fusion, is proposed to fuse the information between lip movement and speech signal. The experimental results indicate that this type of fusion strategy improve the performance of multi-modal person authentication system

In this paper, we present a new speaker diarization system that improves the accuracy of traditional hierarchical clustering-based methods with little increase in computational cost. Our contributions are mainly two fold. First, we include a preprocessing called "local clustering" before the hierarchical clustering algorithm to merge very similar adjacent speech segments. This local clustering aims to reduce the number of segments to be clustered by the hierarchical clustering, so as to dramatically increase the processing speed. Second, we perform a postprocessing called "cross EM refinement" to purify the clusters generated by the hierarchical clustering. This algorithm is based on the idea of cross validation and EM algorithm. Our experimental evaluations show that the proposed cross EM refinement approach reduces the speaker diarization error by up to 56%, with an average reduction of 22% compared to the traditional hierarchical clustering method.

Object tracking plays an important role in multimedia surveillance systems, in which the major types of data are video and audio captured by cameras and microphone arrays. In this paper, we describe a systematic approach to audiovisual object tracking, originally proposed by Beal et al, based on graphical models that jointly combine audio and video variables under a single probabilistic framework. We seek to improve this approach through three aspects: First, we introduce background subtraction preprocessing of video data. Second, we modify the video model to exclude the background from being transformed. Third, we extend the joint model to a dynamic Bayes net. These improvements yield satisfactory results on single person tracking in a noisy outdoor environment with far-field background road traffic, and handle situations where the target is lost due to occlusions

Person-based indices and timelines can enable fast and non-linear access to recorded meetings. This paper focuses on how to automatically construct those indices and timelines by using face recognition techniques. While there exist extensive research in generic face recognition, recognizing faces in recorded meetings is still an understudied area. Real-world meeting videos impose several interesting and unique challenges including complex lighting, low imaging quality, and large variations in head pose and size. In this paper, a promising approach based on MRC-Boosting is presented to address these challenges, which achieves encouraging performance on real-world meeting videos and shows superior accuracy and robustness compared to two popular existing approaches

We present a robust vision system for single person tracking inside a smart room using multiple synchronized, calibrated, stationary cameras. The system consists of two main components, namely initialization and tracking, assisted by an additional component that detects tracking drift. The main novelty lies in the adaptive tracking mechanism that is based on subspace learning of the tracked person appearance in selected two-dimensional camera views. The sub-space is learned on the fly, during tracking, but in contrast to the traditional literature approach, an additional "forgetting" mechanism is introduced, as a means to reduce drifting. The proposed algorithm replaces mean-shift tracking, previously employed in our work. By combining the proposed technique with a robust initialization component that is based on face detection and spatio-temporal dynamic programming, the resulting vision system significantly outperforms previously reported systems for the task of tracking the seminar presenter in data collected as part of the CHIL project

A widely accepted prediction is that computing will move to the background, weaving itself into the fabric of our everyday living spaces and projecting the human user into the foreground. If this prediction is to come true, then next generation computing, which we will call human computing, should be about anticipatory user interfaces that should be human-centered, built for humans based on human models. They should transcend the traditional keyboard and mouse to include natural, human-like interactive functions including understanding and emulating certain human behaviors such as affective and social signaling. This article discusses a number of components of human behavior, how they might be integrated into computers, and how far we are from realizing the front end of human computing, that is, how far are we from enabling computers to understand human behavior.

Automatic multimodal recognition of spontaneous affective expressions is a largely unexplored and challenging problem. In this paper, we explore audio-visual emotion recognition in a realistic human conversation setting - Adult Attachment Interview (AAI). Based on the assumption that facial expression and vocal expression be at the same coarse affective states, positive and negative emotion sequences are labeled according to Facial Action Coding System Emotion Codes. Facial texture in visual channel and prosody in audio channel are integrated in the framework of Adaboost multi-stream hidden Markov model (AMHMM) in which Adaboost learning scheme is used to build component HMM fusion. Our approach is evaluated in the preliminary AAI spontaneous emotion recognition experiments.

A novel framework for text-independent speaker verification is proposed. The framework is based on a new interpretation of Universal Background Model. The UBM in our framework actually defines a transform which maps the variable length observation into a fixed dimensional supervector(supervector space). Each speech utterance is then mapped into a point in this supervector space. The similarity measure in this vector space is progressively refined via an iterative cohort modeling scheme. The experiments on NIST 2002 corpus show the effectiveness of this new framework. Overall the EER drops from the baseline system(with T-Norm) 9.21 % to final improved system(without T-Norm) 8.07%. The new framework can effectively reduce the data dependence in the final output score which is clearly indicated in the second sets of experiments. The EER after T-Norm of final system marginally increases by relatively 1.73 % compared to the EER of baseline system drops 16.12 % relatively after T-Norm. Also, the relative improvement of DCF after T-Norm is marginal for the final improved system (2.47%) compared to 33.68 % in baseline system. It clear shows that the iterative cohort modeling effectively reduce the data dependence of the final scores, so that T-Norm will not further improve the system performance. Also, the performance of novel frame clearly increases as the iteration grows which suggest that the framework progressively refine the similarity measure on the supervector space with the iterative cohort modeling. Index Terms: speaker verification, utterance transform, iterative cohort modeling.

In this paper, we present a spectral clustering approach to explore the possibility of discovering structure from audio data. To apply the Ng-Jordan-Weiss (NJW) spectral clustering algorithm to speaker diarization, we propose some domain specific solutions to the open issues of this algorithm: choice of metric; selection of scaling parameter; estimation of the number of clusters. Then, a postprocessing step – “Cross EM refinement ” – is conducted to further improve the performance of spectral learning. In experiments, this approach has performance very similar to the traditional hierarchical clustering on the audio data of Japanese Parliament Panel Discussions, but it runs much faster than the latter.

In this paper, we present a novel spectral analysis algorithm for image clustering. First, the image manifold is embedded onto a low-dimensional feature space with dual objectives, i.e., maximizing the distances of faraway sample pairs meanwhile preserving the local manifold structure, which essentially results in a Trace Ratio optimization problem. Then an efficient iterative procedure is proposed to directly optimize the trace ratio and finally the clustering process is implemented on the derived low-dimensional embedding. Moreover, the linear approximation is also presented for handling the out-of-sample data. Experimental results show that our algorithm, referred to as Maximum Unfolded Embedding, brings an encouraging improvement in clustering accuracy over the state-of-the-art algorithms, such as K-Means, PCA-Kmeans, normalized cut \cite shi00normalized, and Locality Preserving Clustering [13].

To simulate the human ability to assess affects, an automatic affect recognition system should make use of multi-sensor information. In the framework of multi-stream fused hidden Markov model (MFHMM), we present a training combination strategy towards audio-visual affect recognition. Different from the weighting combination scheme, our approach is able to use a variety of learning methods to obtain a robust multi-stream fusion result. We evaluate our approach in personal-independent recognition of 11 affective states from 20 subjects. The experimental results suggest that MFHMM outperforms IHMM which assumes the independence among streams, and the training combination strategy has the superiority over the weighting combination under clean and varying audio channel noise condition.

We propose an HMM model for contour detection based on multiple visual cues in spatial domain and improve it by joint probabilistic matching to reduce background clutter. It is further integrated with unscented Kalman filter to exploit object dynamics in nonlinear systems for robust contour tracking.

In this paper, we present the logarithmic total variation (LTV) model for face recognition under varying illumination, including natural lighting conditions, where we rarely know the strength, direction, or number of light sources. The proposed LTV model has the ability to factorize a single face image and obtain the illumination invariant facial structure, which is then used for face recognition. Our model is inspired by the SQI model but has better edge-preserving ability and simpler parameter selection. The merit of this model is that neither does it require any lighting assumption nor does it need any training. The LTV model reaches very high recognition rates in the tests using both Yale and CMU PIE face databases as well as a face database containing 765 subjects under outdoor lighting conditions.

This paper presents a new method of motion analysis of articulated objects from feature point correspondences over monocular perspective images without imposing any constraints on motion. An articulated object is modeled as a kinematic chain consisting of joints and links, and its 3D joint positions are estimated within a scale factor using the connection relationship of two links over two or three images. Then, twists and exponential maps are employed to represent the motion of each link, including the general motion of the base link and the rotation of other links around their joints. Finally, constraints from image point correspondences, which are similar to that of the essential matrix in rigid motion, are developed to estimate the motion. In the algorithm, the characteristic of articulated motion, i.e., motion correlation among links, is applied to decrease the complexity of the problem and improve the robustness. A point pattern matching algorithm for articulated objects is also discussed in this paper. Simulations and experiments on real images show the correctness and efficiency of the algorithms.