Diffusion models are proficient at generating high-quality images. They are however effective only when operating at the resolution used during training. Inference at a scaled resolution leads to repetitive patterns and structural distortions. Retraining at higher resolutions quickly becomes prohibitive. Thus, methods enabling pre-existing diffusion models to operate at flexible test-time resolutions are highly desirable. Previous works suffer from frequent artifacts and often introduce large latency overheads. We propose two simple modules that combine to solve these issues. We introduce a Frequency Modulation (FM) module that leverages the Fourier domain to improve the global structure consistency, and an Attention Modulation (AM) module which improves the consistency of local texture patterns, a problem largely ignored in prior works. Our method, coined FAM diffusion, can seamlessly integrate into any latent diffusion model and requires no additional training. Extensive qualitative results highlight the effectiveness of our method in addressing structural and local artifacts, while quantitative results show state-of-the-art performance. Also, our method avoids redundant inference tricks for improved consistency such as patch-based or progressive generation, leading to negligible latency overheads. FAM diffusion project webpage: https://happy-hsy.github.io/projects/Famdiffusion/

There has been immense progress recently in the visual quality of Stable Diffusion-based Super Resolution (SD-SR). However, deploying large diffusion models on computationally restricted devices such as mobile phones remains impractical due to the large model size and high latency. This is compounded for SR as it often operates at high res (e.g. 4K×3K). In this work, we introduce Edge-SD-SR, the first parameter efficient and low latency diffusion model for image super-resolution. Edge-SD-SR consists of ~169M parameters, including UNet, encoder and decoder, and has a complexity of only ∼ 142 GFLOPs. To maintain a high visual quality on such low compute budget, we introduce a number of training strategies: (i) A novel conditioning mechanism on the low-resolution input, coined bidirectional conditioning, which tailors the SD model for the SR task. (ii) Joint training of the UNet and encoder, while decoupling the encodings of the HR and LR images and using a dedicated schedule. (iii) Finetuning the decoder using the UNet’s output to directly tailor the decoder to the latents obtained at inference time. Edge-SD-SR runs efficiently on device, e.g. it can upscale a 128 × 128 patch to 512 × 512 in 38 msec while running on a Samsung S24 DSP, and of a 512 × 512 to 2, 048 × 2, 048 (requiring 25 model evaluations) in just ∼ 1.1 sec. Furthermore, we show that Edge-SD-SR matches or even outperforms state-of-the-art SR approaches on the most established SR benchmarks.

Contrastively-trained Vision-Language Models (VLMs) like CLIP have become the de facto approach for discriminative vision-language representation learning. However, these models have limited language understanding, often exhibiting a "bag of words" behavior. At the same time, Large Vision-Language Models (LVLMs), which combine vision encoders with LLMs, have been shown to be capable of detailed vision-language reasoning, yet their autoregressive nature renders them less suitable for discriminative tasks.In this work, we propose to combine "the best of both worlds": a new training approach for discriminative fine-tuning of LVLMs that results in strong discriminative and compositional capabilities. Essentially, our approach converts a generative LVLM into a discriminative one, unlocking its capability for powerful image-text discrimination combined with enhanced language understanding.Our contributions include (1) A carefully designed training/optimization framework that utilizes image-text pairs of variable length and granularity for training the model with both contrastive and next-token prediction losses. This is accompanied by ablation studies that justify the necessity of our framework’s components. (2) A parameter-efficient adaptation method using a combination of soft prompting and LoRA adapters. (3) Significant improvements over state-of-the-art CLIP-like models of similar size, including standard image-text retrieval benchmarks and notable gains in compositionality.

Abstract Existing knowledge distillation methods mostly focus on distillation of teacher’s prediction and intermediate activation. However, the structured representation, which arguably is one of the most critical ingredients of deep models, is largely overlooked. In this work, we propose a novel semantic representational distillation (SRD) method dedicated for distilling representational knowledge semantically from a pretrained teacher to a target student. The key idea is that we leverage the teacher’s classifier as a semantic critic for evaluating the representations of both teacher and student and distilling the semantic knowledge with high-order structured information over all feature dimensions. This is accomplished by introducing a notion of cross-network logit computed through passing student’s representation into teacher’s classifier. Further, considering the set of seen classes as a basis for the semantic space in a combinatorial perspective, we scale SRD to unseen classes for enabling effective exploitation of largely available, arbitrary unlabeled training data. At the problem level, this establishes an interesting connection between knowledge distillation with open-set semi-supervised learning (SSL). Extensive experiments show that our SRD outperforms significantly previous state-of-the-art knowledge distillation methods on both coarse object classification and fine face recognition tasks, as well as less studied yet practically crucial binary network distillation. Under more realistic open-set SSL settings we introduce, we reveal that knowledge distillation is generally more effective than existing out-of-distribution sample detection, and our proposed SRD is superior over both previous distillation and SSL competitors. The source code is available at https://github.com/jingyang2017/SRD_ossl .

Hai Pham, Isma Hadji, Xinnuo Xu, Ziedune Degutyte, Jay Rainey, Evangelos Kazakos, Afsaneh Fazly, Georgios Tzimiropoulos, Brais Martinez. Proceedings of the 18th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

Current Vision-Language (VL) models owe their success to large-scale pre-training on webcollected data, which in turn requires highcapacity architectures and large compute resources for training.We posit that when the downstream tasks are known in advance, which is in practice common, the pretraining process can be aligned to the downstream domain, leading to more efficient and accurate models, while shortening the pretraining step.To this end, we introduce a domain-aligned pretraining strategy that, without additional data collection, improves the accuracy on a domain of interest, herein, that of human activities, while largely preserving the generalist knowledge.At the core of our approach stands a new LLM-based method that, provided with a simple set of concept seeds, produces a concept hierarchy with high coverage of the target domain.The concept hierarchy is used to filter a large-scale webcrawled dataset and, then, enhance the resulting instances with targeted synthetic labels.We study in depth how to train such approaches and their resulting behavior.We further show generalization to video-based data by introducing a fast adaptation approach for transitioning from a static (image) model to a dynamic one (i.e. with temporal modeling).On the domain of interest, our approach significantly outperforms models trained on up to 60× more samples and between 10 -100× shorter training schedules for image retrieval, video retrieval and action recognition.Code will be released.

Data point selection (DPS) is becoming a critical topic in deep learning due to the ease of acquiring uncurated training data compared to the difficulty of obtaining curated or processed data. Existing approaches to DPS are predominantly based on a bi-level optimisation (BLO) formulation, which is demanding in terms of memory and computation, and exhibits some theoretical defects regarding minibatches. Thus, we propose a novel Bayesian approach to DPS. We view the DPS problem as posterior inference in a novel Bayesian model where the posterior distributions of the instance-wise weights and the main neural network parameters are inferred under a reasonable prior and likelihood model. We employ stochastic gradient Langevin MCMC sampling to learn the main network and instance-wise weights jointly, ensuring convergence even with minibatches. Our update equation is comparable to the widely used SGD and much more efficient than existing BLO-based methods. Through controlled experiments in both the vision and language domains, we present the proof-of-concept. Additionally, we demonstrate that our method scales effectively to large language models and facilitates automated per-task optimization for instruction fine-tuning datasets.

This paper is on Few-Shot Object Detection (FSOD), where given a few templates (examples) depicting a novel class (not seen during training), the goal is to detect all of its occurrences within a set of images. From a practical perspective, an FSOD system must fulfil the following desiderata: (a) it must be used as is, without requiring any fine-tuning at test time, (b) it must be able to process an arbitrary number of novel objects concurrently while supporting an arbitrary number of examples from each class and (c) it must achieve accuracy comparable to a closed system. Towards satisfying (a)-(c), in this work, we make the following contributions: We introduce, for the first time, a simple, yet powerful, few-shot detection transformer (FS-DETR) based on visual prompting that can address both desiderata (a) and (b). Our system builds upon the DETR framework, extending it based on two key ideas: (1) feed the provided visual templates of the novel classes as visual prompts during test time, and (2) "stamp" these prompts with pseudo-class embeddings (akin to soft prompting), which are then predicted at the output of the decoder. Importantly, we show that our system is not only more flexible than existing methods, but also, it makes a step towards satisfying desideratum (c). Specifically, it is significantly more accurate than all methods that do not require fine-tuning and even matches and outperforms the current state-of-the-art fine-tuning based methods on the most well-established benchmarks (PASCAL VOC & MSCOCO).

This paper sets out to solve the following problem: How can we turn a generative video captioning model into an open-world video/action classification model? Video captioning models can naturally produce open-ended free-form descriptions of a given video which, however, might not be discriminative enough for video/action recognition. Unfortunately, when fine-tuned to auto-regress the class names directly, video captioning models overfit the base classes losing their open-world zero-shot capabilities. To alleviate base class overfitting, in this work, we propose to use reinforcement learning to enforce the output of the video captioning model to be more class-level discriminative. Specifically, we propose ReGen, a novel reinforcement learning based framework with a three-fold objective and reward functions: (1) a class-level discrimination reward that enforces the generated caption to be correctly classified into the corresponding action class, (2) a CLIP reward that encourages the generated caption to continue to be descriptive of the input video (i.e. video-specific), and (3) a grammar reward that preserves the grammatical correctness of the caption. We show that ReGen can train a model to produce captions that are: discriminative, video-specific and grammatically correct. Importantly, when evaluated on standard benchmarks for zero- and few-shot action classification, ReGen significantly outperforms the previous state-of-the-art.

Foundational image-language models have generated considerable interest due to their efficient adaptation to downstream tasks by prompt learning. Prompt learning treats part of the language model input as trainable while freezing the rest, and optimizes an Empirical Risk Mini-mization objective. However, Empirical Risk Minimization is known to suffer from distributional shifts which hurt gen-eralizability to prompts unseen during training. By leveraging the regularization ability of Bayesian methods, we frame prompt learning from the Bayesian perspective and formulate it as a variational inference problem. Our approach regularizes the prompt space, reduces overfitting to the seen prompts and improves the prompt generalization on unseen prompts. Our framework is implemented by modeling the input prompt space in a probabilistic manner, as an a priori distribution which makes our proposal compatible with prompt learning approaches that are unconditional or conditional on the image. We demonstrate empirically on 15 benchmarks that Bayesian prompt learning provides an appropriate coverage of the prompt space, prevents learning spurious features, and exploits transferable invariant features. This results in better generalization of unseen prompts, even across different datasets and domains.Code available at: https://github.com/saic-fi/Bayesian-Prompt-Learning

Vision-Language (V-L) models trained with contrastive learning to align the visual and language modalities have been shown to be strong few-shot learners. Soft prompt learning is the method of choice for few-shot downstream adaption aiming to bridge the modality gap caused by the distribution shift induced by the new domain. While parameter-efficient, prompt learning still requires access to the model weights and can be computationally infeasible for large models with billions of parameters. To address these shortcomings, in this work, we describe a black-box method for V-L few-shot adaptation that (a) operates on pre-computed image and text features and hence works without access to the model’s weights, (b) it is orders of magnitude faster at training time, (c) it is amenable to both supervised and unsupervised training, and (d) it can be even used to align image and text features computed from uni-modal models. To achieve this, we propose Linear Feature Alignment (LFA), a simple linear approach for V-L re-alignment in the target domain. LFA is initialized from a closed-form solution to a least-squares problem and then it is iteratively updated by minimizing a re-ranking loss. Despite its simplicity, our approach can even surpass soft-prompt learning methods as shown by extensive experiments on 11 image and 2 video datasets.Code available at: https://github.com/saic-fi/LFA

Lipreading has witnessed a lot of progress due to the resurgence of neural networks. Recent works have placed emphasis on aspects such as improving performance by finding the optimal architecture or improving generalization. However, there is still a significant gap between the current methodologies and the requirements for an effective deployment of lipreading in practical scenarios. In this work, we propose a series of innovations that significantly bridge that gap: first, we raise the state-of-the-art performance by a wide margin on LRW and LRW-1000 to 88.5 % and 46.6 %, respectively using self-distillation. Secondly, we propose a series of architectural changes, including a novel Depthwise Separable Temporal Convolutional Network (DS-TCN) head, that slashes the computational cost to a fraction of the (already quite efficient) original model. Thirdly, we show that knowledge distillation is a very effective tool for recovering performance of the lightweight models. This results in a range of models with different accuracy-efficiency trade-offs. However, our most promising lightweight models are on par with the current state-of-the-art while showing a reduction of 8.2× and 3.9× in terms of computational cost and number of parameters, respectively, which we hope will enable the deployment of lipreading models in practical applications.

Many video analysis tasks require temporal localization for the detection of content changes. However, most existing models developed for these tasks are pre-trained on general video action classification tasks. This is due to large scale annotation of temporal boundaries in untrimmed videos being expensive. Therefore, no suitable datasets exist that enable pre-training in a manner sensitive to temporal boundaries. In this paper for the first time, we investigate model pre-training for temporal localization by introducing a novel boundary-sensitive pretext (BSP) task. Instead of relying on costly manual annotations of temporal boundaries, we propose to synthesize temporal boundaries in existing video action classification datasets. By defining different ways of synthesizing boundaries, BSP can then be simply conducted in a self-supervised manner via the classification of the boundary types. This enables the learning of video representations that are much more transferable to downstream temporal localization tasks. Extensive experiments show that the proposed BSP is superior and complementary to the existing action classification-based pre-training counterpart, and achieves new state-of-the-art performance on several temporal localization tasks. Please visit our website for more details https://frostinassiky.github.io/bsp.

This paper is on video recognition using Transformers. Very recent attempts in this area have demonstrated promising results in terms of recognition accuracy, yet they have been also shown to induce, in many cases, significant computational overheads due to the additional modelling of the temporal information. In this work, we propose a Video Transformer model the complexity of which scales linearly with the number of frames in the video sequence and hence induces no overhead compared to an image-based Transformer model. To achieve this, our model makes two approximations to the full space-time attention used in Video Transformers: (a) It restricts time attention to a local temporal window and capitalizes on the Transformer's depth to obtain full temporal coverage of the video sequence. (b) It uses efficient space-time mixing to attend jointly spatial and temporal locations without inducing any additional cost on top of a spatial-only attention model. We also show how to integrate 2 very lightweight mechanisms for global temporal-only attention which provide additional accuracy improvements at minimal computational cost. We demonstrate that our model produces very high recognition accuracy on the most popular video recognition datasets while at the same time being significantly more efficient than other Video Transformer models.

Most existing temporal action localization (TAL) methods rely on a transfer learning pipeline: by first optimizing a video encoder on a large action classification dataset (i.e., source domain), followed by freezing the encoder and training a TAL head on the action localization dataset (i.e., target domain). This results in a task discrepancy problem for the video encoder – trained for action classification, but used for TAL. Intuitively, joint optimization with both the video encoder and TAL head is a strong baseline solution to this discrepancy. However, this is not operable for TAL subject to the GPU memory constraints, due to the prohibitive computational cost in processing long untrimmed videos. In this paper, we resolve this challenge by introducing a novel low-fidelity (LoFi) video encoder optimization method. Instead of always using the full training configurations in TAL learning, we propose to reduce the mini-batch composition in terms of temporal, spatial, or spatio-temporal resolution so that jointly optimizing the video encoder and TAL head becomes operable under the same memory conditions of a mid-range hardware budget. Crucially, this enables the gradients to flow backwards through the video encoder conditioned on a TAL supervision loss, favourably solving the task discrepancy problem and providing more effective feature representations. Extensive experiments show that the proposed LoFi optimization approach can significantly enhance the performance of existing TAL methods. Encouragingly, even with a lightweight ResNet18 based video encoder in a single RGB stream, our method surpasses two-stream (RGB + optical-flow) ResNet50 based alternatives, often by a good margin. Our code is publicly available at https://github.com/saic-fi/lofiactionlocalization.

Lip-reading has attracted a lot of research attention lately thanks to advances in deep learning. The current state-of-the-art model for recognition of isolated words in-the-wild consists of a residual network and Bidirectional Gated Recurrent Unit (BGRU) layers. In this work, we address the limitations of this model and we propose changes which further improve its performance. Firstly, the BGRU layers are replaced with Temporal Convolutional Networks (TCN). Secondly, we greatly simplify the training procedure, which allows us to train the model in one single stage. Thirdly, we show that the current state-of-the-art methodology produces models that do not generalize well to variations on the sequence length, and we address this issue by proposing a variable-length augmentation. We present results on the largest publicly-available datasets for isolated word recognition in English and Mandarin, LRW and LRW1000, respectively. Our proposed model1results in an absolute improvement of 1.2% and 3.2%, respectively, in these datasets which is the new state-of-the-art performance.

Action recognition has seen a dramatic performance improvement in the last few years. Most of the current state-of-the-art literature either aims at improving performance through changes to the backbone CNN network, or exploring different trade-offs between computational efficiency and performance, again through altering the backbone network. However, almost all of these works maintain the same last layers of the network, which simply consist of a global average pooling followed by a fully connected layer. In this work we focus on how to improve the representation capacity of the network, but rather than altering the backbone, we focus on improving the last layers of the network, where changes have low impact in terms of computational cost. In particular, we hypothesize that current architectures have poor sensitivity to finer details and we exploit recent advances in the fine-grained recognition literature to improve our model in this aspect. With the proposed approach, we obtain state-of-the-art performance on Kinetics-400 and Something-Something-V1, the two major large-scale action recognition benchmarks.

As one of the most comprehensive and objective ways to describe facial expressions, the Facial Action Coding System (FACS) has recently received significant attention. Over the past 30 years, extensive research has been conducted by psychologists and neuroscientists on various aspects of facial expression analysis using FACS. Automating FACS coding would make this research faster and more widely applicable, opening up new avenues to understanding how we communicate through facial expressions. Such an automated process can also potentially increase the reliability, precision and temporal resolution of coding. This paper provides a comprehensive survey of research into machine analysis of facial actions. We systematically review all components of such systems: pre-processing, feature extraction and machine coding of facial actions. In addition, the existing FACS-coded facial expression databases are summarised. Finally, challenges that have to be addressed to make automatic facial action analysis applicable in real-life situations are extensively discussed. There are two underlying motivations for us to write this survey paper: the first is to provide an up-to-date review of the existing literature, and the second is to offer some insights into the future of machine recognition of facial actions: what are the challenges and opportunities that researchers in the field face.

Linear regression is a fundamental building block in many face detection and tracking algorithms, typically used to predict shape displacements from image features through a linear mapping. This paper presents a Functional Regression solution to the least squares problem, which we coin Continuous Regression, resulting in the first real-time incremental face tracker. Contrary to prior work in Functional Regression, in which B-splines or Fourier series were used, we propose to approximate the input space by its first-order Taylor expansion, yielding a closed-form solution for the continuous domain of displacements. We then extend the continuous least squares problem to correlated variables, and demonstrate the generalisation of our approach. We incorporate Continuous Regression into the cascaded regression framework, and show its computational benefits for both training and testing. We then present a fast approach for incremental learning within Cascaded Continuous Regression, coined iCCR, and show that its complexity allows real-time face tracking, being 20 times faster than the state of the art. To the best of our knowledge, this is the first incremental face tracker that is shown to operate in real-time. We show that iCCR achieves state-of-the-art performance on the 300-VW dataset, the most recent, large-scale benchmark for face tracking.

Automatic continuous time, continuous value assessment of a patient's pain from face video is highly sought after by the medical profession. Despite the recent advances in deep learning that attain impressive results in many domains, pain estimation risks not being able to benefit from this due to the difficulty in obtaining data sets of considerable size. In this work we propose a combination of hand-crafted and deep-learned features that makes the most of deep learning techniques in small sample settings. Encoding shape, appearance, and dynamics, our method significantly outperforms the current state of the art, attaining a RMSE error of less than 1 point on a 16-level pain scale, whilst simultaneously scoring a 67.3% Pearson correlation coefficient between our predicted pain level time series and the ground truth.

This paper explores the use of context on regression-based methods for facial landmarking. Regression based methods have revolutionised facial landmarking solutions. In particular those that implicitly infer the whole shape of a structured object have quickly become the state-of-the-art. The most notable exemplar is the Supervised Descent Method (SDM). Its main characteristics are the use of the cascaded regression approach, the use of the full appearance as the inference input, and the aforementioned aim to directly predict the full shape. In this article we argue that the key aspects responsible for the success of SDM are the use of cascaded regression and the avoidance of the constrained optimisation problem that characterised most of the previous approaches. We show that, surprisingly, it is possible to achieve comparable or superior performance using only landmark-specific predictors, which are linearly combined. We reason that augmenting the input with too much context (of which using the full appearance is the extreme case) can be harmful. In fact, we experimentally found that there is a relation between the data variance and the benefits of adding context to the input. We finally devise a simple greedy procedure that makes use of this fact to obtain superior performance to the SDM, while maintaining the simplicity of the algorithm. We show extensive results both for intermediate stages devised to prove the main aspects of the argumentative line, and to validate the overall performance of two models constructed based on these considerations.

Pain-related emotions are a major barrier to effective self rehabilitation in chronic pain. Automated coaching systems capable of detecting these emotions are a potential solution. This paper lays the foundation for the development of such systems by making three contributions. First, through literature reviews, an overview of how pain is expressed in chronic pain and the motivation for detecting it in physical rehabilitation is provided. Second, a fully labelled multimodal dataset (named `EmoPain') containing high resolution multiple-view face videos, head mounted and room audio signals, full body 3D motion capture and electromyographic signals from back muscles is supplied. Natural unconstrained pain related facial expressions and body movement behaviours were elicited from people with chronic pain carrying out physical exercises. Both instructed and non-instructed exercises were considered to reflect traditional scenarios of physiotherapist directed therapy and home-based self-directed therapy. Two sets of labels were assigned: level of pain from facial expressions annotated by eight raters and the occurrence of six pain-related body behaviours segmented by four experts. Third, through exploratory experiments grounded in the data, the factors and challenges in the automated recognition of such expressions and behaviour are described, the paper concludes by discussing potential avenues in the context of these findings also highlighting differences for the two exercise scenarios addressed.

In this article we explore the problem of constructing person-specific models for the detection of facial Action Units (AUs), addressing the problem from the point of view of Transfer Learning and Multi-Task Learning. Our starting point is the fact that some expressions, such as smiles, are very easily elicited, annotated, and automatically detected, while others are much harder to elicit and to annotate. We thus consider a novel problem: all AU models for the target subject are to be learnt using person-specific annotated data for a reference AU (AU12 in our case), and no data or little data regarding the target AU. In order to design such a model, we propose a novel Multi-Task Learning and the associated Transfer Learning framework, in which we consider both relations across subjects and AUs. That is to say, we consider a tensor structure among the tasks. Our approach hinges on learning the latent relations among tasks using one single reference AU, and then transferring these latent relations to other AUs. We show that we are able to effectively make use of the annotated data for AU12 when learning other person-specific AU models, even in the absence of data for the target task. Finally, we show the excellent performance of our method when small amounts of annotated data for the target tasks are made available.

This paper proposes a novel approach to part-based tracking by replacing local matching of an appearance model by direct prediction of the displacement between local image patches and part locations. We propose to use cascaded regression with incremental learning to track generic objects without any prior knowledge of an object's structure or appearance. We exploit the spatial constraints between parts by implicitly learning the shape and deformation parameters of the object in an online fashion. We integrate a multiple temporal scale motion model to initialise our cascaded regression search close to the target and to allow it to cope with occlusions. Experimental results show that our tracker ranks first on the CVPR 2013 Benchmark.

In this paper we propose a new method for the detection of action units that relies on a novel region-based face representation and a mid-level decision layer that combines region-specific information. Different from other approaches, we do not represent the face as a regular grid based on the face location alone (holistic representation), nor by using small patches centred at iducial facial point locations (local representation). Instead, we propose to use domain knowledge regarding AU-specific facial muscle contractions to define a set of face regions covering the whole face. Therefore, as opposed to local appearance models, our face representation makes use of the full facial appearance, while the use of facial point locations to define the regions means that we obtain better-registered descriptors compared to holistic representations. Finally, we propose an AU-specific weighted sum model is used as a decision-level fusion layer in charge of combining region-specific probabilistic information. This configuration allows each classier to learning the typical appearance changes for a specific face part and reduces the dimensionality of the problem thus proving to be more robust. Our approach is evaluated on the DISFA and GEMEP-FERA datasets using two histogram-based appearance features, Local Binary Pattern and Local Phase Quantisation. We show superior performance for both the domain-specific region definition and the decision-level fusion respect to the standard approaches when it comes to automatic facial action unit detection.

Both the configuration and the dynamics of facial expressions are crucial for the interpretation of human facial behavior. Yet to date, the vast majority of reported efforts in the field either do not take the dynamics of facial expressions into account, or focus only on prototypic facial expressions of six basic emotions. Facial dynamics can be explicitly analyzed by detecting the constituent temporal segments in Facial Action Coding System (FACS) Action Units (AUs)-onset, apex, and offset. In this paper, we present a novel approach to explicit analysis of temporal dynamics of facial actions using the dynamic appearance descriptor Local Phase Quantization from Three Orthogonal Planes (LPQ-TOP). Temporal segments are detected by combining a discriminative classifier for detecting the temporal segments on a frame-by-frame basis with Markov Models that enforce temporal consistency over the whole episode. The system is evaluated in detail over the MMI facial expression database, the UNBC-McMaster pain database, the SAL database, the GEMEP-FERA dataset in database-dependent experiments, in cross-database experiments using the Cohn-Kanade, and the SEMAINE databases. The comparison with other state-of-the-art methods shows that the proposed LPQ-TOP method outperforms the other approaches for the problem of AU temporal segment detection, and that overall AU activation detection benefits from dynamic appearance information.

We propose a new algorithm to detect facial points in frontal and near-frontal face images. It combines a regression-based approach with a probabilistic graphical model-based face shape model that restricts the search to anthropomorphically consistent regions. While most regression-based approaches perform a sequential approximation of the target location, our algorithm detects the target location by aggregating the estimates obtained from stochastically selected local appearance information into a single robust prediction. The underlying assumption is that by aggregating the different estimates, their errors will cancel out as long as the regressor inputs are uncorrelated. Once this new perspective is adopted, the problem is reformulated as how to optimally select the test locations over which the regressors are evaluated. We propose to extend the regression-based model to provide a quality measure of each prediction, and use the shape model to restrict and correct the sampling region. Our approach combines the low computational cost typical of regression-based approaches with the robustness of exhaustive-search approaches. The proposed algorithm was tested on over 7,500 images from five databases. Results showed significant improvement over the current state of the art.

In this paper, we present a framework for real-time mosaicing from video sequences recorded from an uncalibrated pan tilt zoom camera based on multiframe registration. To this end, a new frame alignment algorithm, the direct local indirect global (DLIG), is presented. The key idea of the DLIG alignment is to divide the frame alignment problem into the problem of registering a set of spatially related image patches. The registration is iteratively computed by sequentially imposing a good local match and global spatial coherence. The patch registration is performed using a tracking algorithm, so a very efficient local matching can be achieved. We use the patch-based registration to obtain multiframe registration, using the mosaic coordinates to relate the current frame to patches from different frames that partially share the current field of view. Multiframe registration prevents the error accumulation problem, one of the most important problems in mosaicing. We also show how to embed a kernel tracking algorithm in order to obtain a precise and extremely efficient mosaicing algorithm. Finally, we perform a quantitative evaluation of our algorithm, including a comparison with other alignment approaches, and studying its performance against interlaced videos and illumination changes.

Finding fiducial facial points in any frame of a video showing rich naturalistic facial behaviour is an unsolved problem. Yet this is a crucial step for geometric-feature-based facial expression analysis, and methods that use appearance-based features extracted at fiducial facial point locations. In this paper we present a method based on a combination of Support Vector Regression and Markov Random Fields to drastically reduce the time needed to search for a point's location and increase the accuracy and robustness of the algorithm. Using Markov Random Fields allows us to constrain the search space by exploiting the constellations that facial points can form. The regressors on the other hand learn a mapping between the appearance of the area surrounding a point and the positions of these points, which makes detection of the points very fast and can make the algorithm robust to variations of appearance due to facial expression and moderate changes in head pose. The proposed point detection algorithm was tested on 1855 images, the results of which showed we outperform current state of the art point detectors.

In tracking tasks, representing a target region as a weighted histogram has opened possibilities which led to excellent results, as mean shift or camshift algorithms. This representation is extracted from the image by giving weights with kernels and it depends on the properties of the kernels. By a first order Taylor approximation of the histograms it is possible to perform a tracking using several kernels, interpreted as different sources of information. This representation improves the possibilities and gives more flexibility when facing problems of tracking, as occlusions, model variance or projective deformations of the image. In this paper we use this multi-kernel model representation to perform a simultaneous tracking of the entire object and also of each different part individually. This is performed in a new two-step process. In the first step we perform the multikernel estimation and in a second step we update the model representation taking into account single kernel estimations, representing the local movement of each part. From a probabilistic view of the Matusita metric we analyze the usefulness of this method against partial occlusions and some projective transformations like zooms or 3D rotations and articulated movements.