Controllable person image generation aims to generate a person image conditioned on reference images, allowing precise control over the person’s appearance or pose. However, prior methods often distort fine-grained details from the reference image, despite achieving high overall image quality. We attribute these distortions to inadequate attention to corresponding regions in the reference image. To address this, we thereby propose learning flow fields in attention (Leffa), which explicitly guides the target query to attend to the correct reference key in the attention layer during training. Specifically, it is realized via a regularization loss on top of the attention map within a diffusionbased baseline. Our extensive experiments show that Leffa achieves state-of-the-art performance in controlling appearance and pose, significantly reducing fine-grained detail distortion while maintaining high image quality. Additionally, we show that our loss is model-agnostic and can be used to improve the performance of other diffusion models.

We propose SketchINR, to advance the representation of vector sketches with implicit neural models. A variable length vector sketch is compressed into a latent space of fixed dimension that implicitly encodes the underlying shape as a function of time and strokes. The learned function predicts the xy point coordinates in a sketch at each time and stroke. Despite its simplicity, SketchINR outperforms existing representations at multiple tasks: (i) Encoding an entire sketch dataset into a fixed size latent vector, SketchINR gives 60× and 10× data compression over raster and vector sketches, respectively. (ii) SketchINR's auto-decoder provides a much higher-fidelity representation than other learned vector sketch representations, and is uniquely able to scale to complex vector sketches such as FS-COCO. (iii) SketchINR supports parallelisation that can decode/render ∼100× faster than other learned vector representations such as SketchRNN. (iv) SketchINR, for the first time, emulates the human ability to reproduce a sketch with varying abstraction in terms of number and complexity of strokes. As a first look at implicit sketches, SketchINR's compact high-fidelity representation will support future work in modelling long and complex sketches.

In this paper, we explore the unique modality of sketch for explainability, emphasising the profound impact of human strokes compared to conventional pixel-oriented studies. Beyond explanations of network behavior, we discern the genuine implications of explainability across diverse downstream sketch-related tasks. We propose a lightweight and portable explainability solution - a seamless plugin that integrates effortlessly with any pre-trained model, elim-inating the need for re-training. Demonstrating its adaptability, we present four applications: highly studied re-trieval and generation, and completely novel assisted drawing and sketch adversarial attacks. The centrepiece to our solution is a stroke-level attribution map that takes different forms when linked with downstream tasks. By addressing the inherent non-differentiability of rasterisation, we enable explanations at both coarse stroke level (SLA) and partial stroke level (P-SLA), each with its advantages for specific downstream tasks.

In this paper, we democratise 3D content creation, enabling precise generation of 3D shapes from abstract sketches while overcoming limitations tied to drawing skills. We introduce a novel part-level modelling and alignment framework that facilitates abstraction modelling and cross-modal correspondence. Leveraging the same part-level decoder, our approach seamlessly extends to sketch modelling by establishing correspondence between CLIPasso edgemaps and projected 3D part regions, eliminating the need for a dataset pairing human sketches and 3D shapes. Additionally, our method introduces a seamless in-position editing process as a byproduct of cross-modal part-aligned modelling. Operating in a low-dimensional implicit space, our approach significantly reduces computational demands and processing time.

In this study, we explore Transformer-based diffusion models for image and video generation. Despite the dominance of Transformer architectures in various fields due to their flexibility and scalability, the visual generative domain primarily utilizes CNN-based U-Net architectures, particularly in diffusion-based models. We introduce GenTron, a family of Generative models employing Transformer-based diffusion, to address this gap. Our initial step was to adapt Diffusion Transformers (DiTs) from class to text conditioning, a process involving thorough empirical exploration of the conditioning mechanism. We then scale GenTron from approximately 900M to over 3B parameters, observing improvements in visual quality. Furthermore, we extend GenTron to text-to-video generation, incorporating novel motion-free guidance to enhance video quality. In human evaluations against SDXL, GenTron achieves a 51.1% win rate in visual quality (with a 19.8% draw rate), and a 42.3% win rate in text alignment (with a 42.9% draw rate). GenTron notably performs well in T2I-CompBench, highlighting its compositional generation ability. We hope GenTron could provide meaningful insights and serve as a valuable reference for future research. Please refer to the website11https://www.shoufachen.com/gentron_website/ and the arXiv version for the most up-to-date results: https://arxiv.org/abs/2312.04557.

Two primary input modalities prevail in image retrieval: sketch and text. While text is widely used for inter-category retrieval tasks, sketches have been established as the sole preferred modality for fine-grained image retrieval due to their ability to capture intricate visual details. In this paper, we question the reliance on sketches alone for fine-grained image retrieval by simultaneously exploring the fine-grained representation capabilities of both sketch and text, orchestrating a duet between the two. The end result enables precise retrievals previously unattainable, allowing users to pose ever-finer queries and incorporate attributes like colour and contextual cues from text. For this purpose, we introduce a novel compositionality framework, effectively combining sketches and text using pre-trained CLIP models, while eliminating the need for extensive fine-grained textual descriptions. Last but not least, our system extends to novel applications in composed image retrieval, domain attribute transfer, and fine-grained generation, providing solutions for various real-world scenarios.

This paper, for the first time, explores text-to-image diffusion models for Zero-Shot Sketch-based Image Retrieval (ZS-SBIR). We highlight a pivotal discovery: the capacity of text-to-image diffusion models to seamlessly bridge the gap between sketches and photos. This proficiency is underpinned by their robust cross-modal capabilities and shape bias, findings that are substantiated through our pilot studies. In order to harness pre-trained diffusion models effectively, we introduce a straightforward yet powerful strategy focused on two key aspects: selecting optimal feature layers and utilising visual and textual prompts. For the former, we identify which layers are most enriched with information and are best suited for the specific retrieval requirements (category-level or fine-grained). Then we employ visual and textual prompts to guide the model's feature extraction process, enabling it to generate more discriminative and contextually relevant cross-modal representations. Extensive experiments on several benchmark datasets validate significant performance improvements.

In this paper, we propose a novel abstraction-aware sketch-based image retrieval framework capable of handling sketch abstraction at varied levels. Prior works had mainly focused on tackling sub-factors such as drawing style and order, we instead attempt to model abstraction as a whole, and propose feature-level and retrieval granularity-level designs so that the system builds into its DNA the necessary means to interpret abstraction. On learning abstraction-aware features, we for the first-time harness the rich semantic embedding of pre-trained Style-GAN model, together with a novel abstraction-level mapper that deciphers the level of abstraction and dynamically selects appropriate dimensions in the feature matrix correspondingly, to construct a feature matrix embedding that can be freely traversed to accommodate different levels of abstraction. For granularity-level abstraction understanding, we dictate that the retrieval model should not treat all abstraction-levels equally and introduce a differentiable surrogate Acc. @q loss to inject that understanding into the system. Different to the gold-standard triplet loss, our Acc.@q loss uniquely allows a sketch to narrow/broaden its focus in terms of how stringent the evaluation should be - the more abstract a sketch, the less stringent (higher q). Extensive experiments depict our method to outperform existing state-of-the-arts in standard SBIR tasks along with challenging scenarios like early retrieval, forensic sketch-photo matching, and style-invariant retrieval.

This paper unravels the potential of sketches for diffusion models, addressing the deceptive promise of direct sketch control in generative AI. We importantly democratise the process, enabling amateur sketches to generate precise images, living up to the commitment of “what you sketch is what you get”. A pilot study underscores the necessity, revealing that deformities in existing models stem from spatial-conditioning. To rectify this, we propose an abstraction-aware framework, utilising a sketch adapter, adaptive time-step sampling, and discriminative guidance from a pre-trained fine-grained sketch-based image retrieval model, working synergistically to reinforce fine-grained sketch-photo association. Our approach operates seamlessly during inference without the need for textual prompts; a simple, rough sketch akin to what you and I can create suffices! We welcome everyone to examine results presented in the paper and its supplementary. Contributions include democratising sketch control, introducing an abstraction-aware framework, and leveraging discriminative guidance, validated through extensive experiments.

Creating multi-view wire art (MVWA), a static 3D sculpture with diverse interpretations from different viewpoints, is a complex task even for skilled artists. In response, we present DreamWire, an AI system enabling everyone to craft MVWA easily. Users express their vision through text prompts or scribbles, freeing them from intricate 3D wire organisation. Our approach synergises 3D Bézier curves, Prim's algorithm, and knowledge distillation from diffusion models or their variants (e.g., ControlNet). This blend enables the system to represent 3D wire art, ensuring spatial continuity and overcoming data scarcity. Extensive evaluation and analysis are conducted to shed insight on the inner workings of the proposed system, including the trade-off between connectivity and visual aesthetics.

Diffusion models generate images with an unprece-dented level of quality, but how can we freely rearrange image layouts? Recent works generate controllable scenes via learning spatially disentangled latent codes, but these methods do not apply to diffusion models due to their fixed forward process. In this work, we propose SceneDiffusion to optimize a layered scene representation during the diffusion sampling process. Our key insight is that spatial disentanglement can be obtained by jointly denoising scene rende rings at different spatial layouts. Our generated scenes support a wide range of spatial editing operations, including moving, resizing, cloning, and layer-wise appearance editing operations, including object restyling and replacing. Moreover, a scene can be generated conditioned on a ref-erence image, thus enabling object moving for in-the- wild images. Notably, this approach is training-free, compatible with general text-to-image diffusion models, and responsive in less than a second.

Text-to-video editing aims to edit the visual appearance of a source video conditional on textual prompts. A major challenge in this task is to ensure that all frames in the edited video are visually consistent. Most recent works apply advanced text-to-image diffusion models to this task by inflating 2D spatial attention in the U-Net into spatio-temporal attention. Although temporal context can be added through spatio-temporal attention, it may introduce some irrelevant information for each patch and therefore cause inconsistency in the edited video. In this paper, for the first time, we introduce optical flow into the attention module in diffusion model's U-Net to address the inconsistency issue for text-to-video editing. Our method, FLATTEN, enforces the patches on the same flow path across different frames to attend to each other in the attention module, thus improving the visual consistency in the edited videos. Additionally, our method is training-free and can be seamlessly integrated into any diffusion based text-to-video editing methods and improve their visual consistency. Experiment results on existing text-to-video editing benchmarks show that our proposed method achieves the new state-of-the-art performance. In particular, our method excels in maintaining the visual consistency in the edited videos.

Human sketch has already proved its worth in various visual understanding tasks (e.g., retrieval, segmentation, image-captioning, etc). In this paper, we reveal a new trait of sketches – that they are also salient. This is intuitive as sketching is a natural attentive process at its core. More specifically, we aim to study how sketches can be used as a weak label to detect salient objects present in an image. To this end, we propose a novel method that emphasises on how “salient object” could be explained by hand-drawn sketches. To accomplish this, we introduce a photo-to-sketch generation model that aims to generate sequential sketch coordinates corresponding to a given visual photo through a 2D attention mechanism. Attention maps accumulated across the time steps give rise to salient regions in the process. Extensive quantitative and qualitative experiments prove our hypothesis and delineate how our sketch-based saliency detection model gives a competitive performance compared to the state-of-the-art.

In this paper, we extend scene understanding to include that of human sketch. The result is a complete trilogy of scene representation from three diverse and complementary modalities – sketch, photo, and text. Instead of learning a rigid three-way embedding and be done with it, wefocus on learning a flexible joint embedding that fully supports the “optionality” that this complementarity brings. Our embedding supports optionality on two axes: (i) optionality across modalities – use any combination of modalities as query for downstream tasks like retrieval, (ii) optionality across tasks – simultaneously utilising the embedding for either discriminative (e.g., retrieval) or generative tasks (e.g., captioning). This provides flexibility to end-users by exploiting the best of each modality, therefore serving the very purpose behind our proposal of a trilogy in the first place. First, a combination of information-bottleneck and conditional invertible neural networks disentangle the modality-specific component from modality-agnostic in sketch, photo, and text. Second, the modality-agnostic instances from sketch, photo, and text are synergised using a modified cross-attention. Once learned, we show our embedding can accommodate a multi-facet of scene-related tasks, including those enabled for the first time by the inclusion of sketch, all without any task-specific modifications. Project Page: https://pinakinathc.github.io/scenetrilogy

Sketches are highly expressive, inherently capturing subjective and fine-grained visual cues. The exploration of such innate properties of human sketches has, however, been limited to that of image retrieval. In this paper, for the first time, we cultivate the expressiveness of sketches but for the fundamental vision task of object detection. The end result is a sketch-enabled object detection framework that detects based on what you sketch – that “zebra” (e.g., one that is eating the grass) in a herd of zebras (instance-aware detection), and only the part (e.g., “head” of a “zebra”) that you desire (part-aware detection). We further dictate that our model works without (i) knowing which category to expect at testing (zero-shot) and (ii) not requiring additional bounding boxes (as per fully supervised) and class labels (as per weakly supervised). Instead of devising a model from the ground up, we show an intuitive synergy between foundation models (e.g., CLIP) and existing sketch models build for sketch-based image retrieval (SBIR), which can already elegantly solve the task – CLIP to provide model generalisation, and SBIR to bridge the (sketch→photo) gap. In particular, we first perform independent prompting on both sketch and photo branches of an SBIR model to build highly generalisable sketch and photo encoders on the back of the generalisation ability of CLIP. We then devise a training paradigm to adapt the learned encoders for object detection, such that the region embeddings of detected boxes are aligned with the sketch and photo embeddings from SBIR. Evaluating our framework on standard object detection datasets like PASCAL-VOC and MS-COCO outperforms both supervised (SOD) and weakly-supervised object detectors (WSOD) on zero-shot setups. Project Page: https://pinakinathc.github.io/sketch-detect

In the fashion domain, there exists a variety of vision-and-language (V+L) tasks, including cross-modal retrieval, text-guided image retrieval, multi-modal classification, and image captioning. They differ drastically in each individual input/output format and dataset size. It has been common to design a task-specific model and fine-tune it independently from a pre-trained V+l model (e.g., CLIP). This results in parameter inefficiency and inability to exploit inter-task relatedness. To address such issues, we propose a novel FAshion-focused Multi-task Efficient learning method for Vision-and-Language tasks (FAME-ViL) in this work. Compared with existing approaches, FAME-ViL applies a single model for multiple heterogeneous fashion tasks, therefore being much more parameter-efficient. It is enabled by two novel components: (1) a task-versatile architecture with cross-attention adapters and task-specific adapters integrated into a unified V+L model, and (2) a stable and effective multi-task training strategy that supports learning from heterogeneous data and prevents negative transfer. Extensive experiments on four fashion tasks show that our FAME-ViL can save 61.5% of parameters over alternatives, while significantly outperforming the conventional independently trained single-task models. Code is available at https://github.com/BrandonHanx/FAME-ViL.

Given an abstract, deformed, ordinary sketch from untrained amateurs like you and me, this paper turns it into a photorealistic image – just like those shown in Fig. 1(a), all non-cherry-picked. We differ significantly from prior art in that we do not dictate an edgemap-like sketch to start with, but aim to work with abstract free-hand human sketches. In doing so, we essentially democratise the sketch-to-photo pipeline, “picturing” a sketch regardless of how good you sketch. Our contribution at the outset is a decoupled encoder-decoder training paradigm, where the decoder is a StyleGAN trained on photos only. This importantly ensures that generated results are always photorealistic. The rest is then all centred around how best to deal with the abstraction gap between sketch and photo. For that, we propose an autoregressive sketch mapper trained on sketch-photo pairs that maps a sketch to the StyleGAN latent space. We further introduce specific designs to tackle the abstract nature of human sketches, including a fine-grained discriminative loss on the back of a trained sketch-photo retrieval model, and a partial-aware sketch augmentation strategy. Finally, we showcase a few downstream tasks our generation model enables, amongst them is showing how fine-grained sketch-based image retrieval, a well-studied problem in the sketch community, can be reduced to an image (generated) to image retrieval task, surpassing state-of-the-arts. We put forward generated results in the supplementary for everyone to scrutinise. Project page: https://subhadeepkoley.github.io/PictureThatSketch

Existing Temporal Action Detection (TAD) methods typically take a pre-processing step in converting an input varying-length video into a fixed-length snippet representation sequence, before temporal boundary estimation and action classification. This pre-processing step would temporally downsample the video, reducing the inference resolution and hampering the detection performance in the original temporal resolution. In essence, this is due to a temporal quantization error introduced during resolution downsampling and recovery. This could negatively impact the TAD performance, but is largely ignored by existing methods. To address this problem, in this work we introduce a novel model-agnostic post-processing method without model redesign and retraining. Specifically, we model the start and end points of action instances with a Gaussian distribution for enabling temporal boundary inference at a sub-snippet level. We further introduce an efficient Taylor-expansion based approximation, dubbed as Gaussian Approximated Post-processing (GAP). Extensive experiments demonstrate that our GAP can consistently improve a wide variety of pre-trained off-the-shelf TAD models on the challenging ActivityNet (+0.2%~0. 7% in average mAP) and THUMOS (+0.2%~0.5% in average mAP) benchmarks. Such performance gains are already significant and highly comparable to those achieved by novel model designs. Also, GAP can be integrated with model training for further performance gain. Importantly, GAP enables lower temporal resolutions for more efficient inference, facilitating low-resource application. The code is available at https://github.com/sauradip/GAP

This paper, for the very first time, introduces human sketches to the landscape of XAI (Explainable Artificial Intelligence). We argue that sketch as a “human-centred” data form, represents a natural interface to study explainability. We focus on cultivating sketch-specific explainability designs. This starts by identifying strokes as a unique building block that offers a degree of flexibility in object construction and manipulation impossible in photos. Following this, we design a simple explainability-friendly sketch encoder that accommodates the intrinsic properties of strokes: shape, location, and order. We then move on to define the first ever XAI task for sketch, that of stroke location inversion (SLI). Just as we have heat maps for photos, and correlation matrices for text, SLI offers an explainability angle to sketch in terms of asking a network how well it can recover stroke locations of an unseen sketch. We offer qualitative results for readers to interpret as snapshots of the SLI process in the paper, and as GIFs on the project page. A minor but interesting note is that thanks to its sketch-specific design, our sketch encoder also yields the best sketch recognition accuracy to date while having the smallest number of parameters. The code is available at https://sketchxai.github.io.

In this paper, we leverage CLIP for zero-shot sketch based image retrieval (ZS-SBIR). We are largely inspired by recent advances on foundation models and the unparalleled generalisation ability they seem to offer, but for the first time tailor it to benefit the sketch community. We put forward novel designs on how best to achieve this synergy, for both the category setting and the fine-grained setting ('all”}. At the very core of our solution is a prompt learning setup. First we show just via factoring in sketch-specific prompts, we already have a category-level ZS-SBIR system that over-shoots all prior arts, by a large margin (24.8%) - a great testimony on studying the CLIP and ZS-SBIR synergy. Moving onto the fine-grained setup is however trickier, and re-quires a deeper dive into this synergy. For that, we come up with two specific designs to tackle the fine-grained matching nature of the problem: (i) an additional regularisation loss to ensure the relative separation between sketches and photos is uniform across categories, which is not the case for the gold standard standalone triplet loss, and (ii) a clever patch shuffling technique to help establishing instance-level structural correspondences between sketch-photo pairs. With these designs, we again observe signifi-cant performance gains in the region of 26.9% over previ-ous state-of-the-art. The take-home message, if any, is the proposed CLIP and prompt learning paradigm carries great promise in tackling other sketch-related tasks (not limited to ZS-SBIR) where data scarcity remains a great challenge. Project page: https://aneeshan95.github.ioISketchLVM/

This paper advances the fine-grained sketch-based image retrieval (FG-SBIR) literature by putting forward a strong baseline that overshoots prior state-of-the-arts by ≈11 %. This is not via complicated design though, but by addressing two critical issues facing the community (i) the gold standard triplet loss does not enforce holistic latent space geometry, and (ii) there are never enough sketches to train a high accuracy model. For the former, we propose a simple modification to the standard triplet loss, that explicitly enforces separation amongst photos/sketch instances. For the latter, we put forward a novel knowledge distillation module can leverage photo data for model training. Both modules are then plugged into a novel plug-n-playable training paradigm that allows for more stable training. More specifically, for (i) we employ an intra-modal triplet loss amongst sketches to bring sketches of the same instance closer from others, and one more amongst photos to push away different photo instances while bringing closer a structurally augmented version of the same photo (offering a gain of ≈4-6%). To tackle (ii), we first pre-train a teacher on the large set of unlabelled photos over the aforementioned intra-modal photo triplet loss. Then we distill the contextual similarity present amongst the instances in the teacher's embedding space to that in the student's embedding space, by matching the distribution over inter-feature distances of respective samples in both embedding spaces (delivering a further gain of ≈ 4-5%). Apart from outperforming prior arts significantly, our model also yields satisfactory results on generalising to new classes. Project page: https://aneeshan95.github.io/Sketch_PVT/

This paper deals with the problem of localizing objects in image and video datasets from visual exemplars. In particular, we focus on the challenging problem of egocentric visual query localization. We first identify grave implicit biases in current query-conditioned model design and visual query datasets. Then, we directly tackle such biases at both frame and object set levels. Concretely, our method solves these issues by expanding limited annotations and dynamically dropping object proposals during training. Additionally, we propose a novel transformer-based module that allows for object-proposal set context to be considered while incorporating query information. We name our module Conditioned Contextual Transformer or CocoFormer. Our experiments show the proposed adaptations improve egocentric query detection, leading to a better visual query localization system in both 2D and 3D configurations. Thus, we can improve frame-level detection performance from 26.28% to 31.26% in AP, which correspondingly improves the VQ2D and VQ3D localization scores by significant margins. Our improved context-aware query object detector ranked first and second respectively in the VQ2D and VQ3D tasks in the 2nd Ego4D challenge. In addition to this, we showcase the relevance of our proposed model in the Few-Shot Detection (FSD) task, where we also achieve SOTA results. Our code is available at https://github.com/facebookresearch/vq2d_cvpr.

This paper studies the problem of 2D sketch to 3D shape retrieval, but with a focus on democratising the process. We would like this democratisation to happen on two fronts: (i) to remove the need for large-scale specifically sourced 2D sketch and 3D shape datasets, and (ii) to remove restrictions on how well the user needs to sketch and from what viewpoints. The end result is a system that is trainable using existing datasets, and once trained allows users to sketch regardless of drawing skills and without restriction on view angle. We achieve all this via a clever use of pivoting, along with novel designs that injects 3D understanding of 2D sketches into the system. We perform pivoting using two existing datasets, each from a distant research domain to the other: 2D sketch and photo pairs from the sketch-based image retrieval field (SBIR), and 3D shapes from ShapeNet. It follows that the actual feature pivoting happens on photos from the former and 2D projections from the latter. Doing this already achieves most of our democratisation challenge – the level of 2D sketch abstraction embedded in SBIR dataset offers demoralization on drawing quality, and the whole thing works without a specifically sourced 2D sketch and 3D model pair. To further achieve democratisation on sketching viewpoint, we “lift” 2D sketches to 3D space using Blind Perspective-n-Points (BPnP) that injects 3D-aware information into the sketch encoder. Results show ours achieves competitive performance compared with fully-supervised baselines, while meeting all set democratisation goals.

Controllable person image synthesis aims at rendering a source image based on user-specified changes in body pose or appearance. Prior art approaches leverage pixel-level denoising diffusion models conditioned on the coarse skeleton via cross-attention. This leads to two limitations: low efficiency and inaccurate condition information. To address both issues, a novel Pose-Constrained Latent Diffusion model (PoCoLD) is introduced. Rather than using the skeleton as a sparse pose representation, we exploit DensePose which offers much richer body structure information. To effectively capitalize DensePose at a low cost, we propose an efficient pose-constrained attention module that is capable of modeling the complex interplay between appearance and pose. Extensive experiments show that our PoCoLD outperforms the state-of-the-art competitors in image synthesis fidelity. Critically, it runs 2× faster and consumes 3.6× smaller memory than the latest diffusion-model-based alternative during inference. Our code and models are available at https://github.com/BrandonHanx/PoCoLD.

3D shape modeling is labor-intensive, time-consuming, and requires years of expertise. To facilitate 3D shape modeling, we propose a 3D shape generation network that takes a 3D VR sketch as a condition. We assume that sketches are created by novices without art training and aim to reconstruct geometrically realistic 3D shapes of a given category. To handle potential sketch ambiguity, our method creates multiple 3D shapes that align with the original sketch’s structure. We carefully design our method, training the model step-by-step and leveraging multi-modal 3D shape representation to support training with limited training data. To guarantee the realism of generated 3D shapes we leverage the normalizing flow that models the distribution of the latent space of 3D shapes. To encourage the fidelity of the generated 3D shapes to an input sketch, we propose a dedicated loss that we deploy at different stages of the training process. The code is available at https://github.com/Rowl1ng/3Dsketch2shape.

We propose a new formulation of temporal action detection (TAD) with denoising diffusion, DiffTAD in short. Taking as input random temporal proposals, it can yield action proposals accurately given an untrimmed long video. This presents a generative modeling perspective, against previous discriminative learning manners. This capability is achieved by first diffusing the ground-truth proposals to random ones (i.e., the forward/noising process) and then learning to reverse the noising process (i.e., the backward/denoising process). Concretely, we establish the denoising process in the Transformer decoder (e.g., DETR) by introducing a temporal location query design with faster convergence in training. We further propose a cross-step selective conditioning algorithm for inference acceleration. Extensive evaluations on ActivityNet and THUMOS show that our DiffTAD achieves top performance compared to previous art alternatives. The code is available at https://github.com/sauradip/DiffusionTAD.

Recently, text-guided 3D generative methods have made remarkable advancements in producing high-quality textures and geometry, capitalizing on the proliferation of large vision-language and image diffusion models. However, existing methods still struggle to create high-fidelity 3D head avatars in two aspects: (1) They rely mostly on a pre-trained text-to-image diffusion model whilst missing the necessary 3D awareness and head priors. This makes them prone to inconsistency and geometric distortions in the generated avatars. (2) They fall short in fine-grained editing. This is primarily due to the inherited limitations from the pre-trained 2D image diffusion models, which become more pronounced when it comes to 3D head avatars. In this work, we address these challenges by introducing a versatile coarse-to-fine pipeline dubbed HeadSculpt for crafting (i.e., generating and editing) 3D head avatars from textual prompts. Specifically, we first equip the diffusion model with 3D awareness by leveraging landmark-based control and a learned textual embedding representing the back view appearance of heads, enabling 3D-consistent head avatar generations. We further propose a novel identity-aware editing score distillation strategy to optimize a textured mesh with a high-resolution differentiable rendering technique. This enables identity preservation while following the editing instruction. We showcase HeadSculpt's superior fidelity and editing capabilities through comprehensive experiments and comparisons with existing methods.

Free-hand sketches are highly illustrative, and have been widely used by humans to depict objects or stories from ancient times to the present. The recent prevalence of touchscreen devices has made sketch creation a much easier task than ever and consequently made sketch-oriented applications increasingly popular. The progress of deep learning has immensely benefited free-hand sketch research and applications. This paper presents a comprehensive survey of the deep learning techniques oriented at free-hand sketch data, and the applications that they enable. The main contents of this survey include: (i) A discussion of the intrinsic traits and unique challenges of free-hand sketch, to highlight the essential differences between sketch data and other data modalities, e.g., natural photos. (ii) A review of the developments of free-hand sketch research in the deep learning era, by surveying existing datasets, research topics, and the state-of-the-art methods through a detailed taxonomy and experimental evaluation. (iii) Promotion of future work via a discussion of bottlenecks, open problems, and potential research directions for the community.

Generalization to out-of-distribution (OOD) data is a capability natural to humans yet challenging for machines to reproduce. This is because most learning algorithms strongly rely on the i.i.d. assumption on source/target data, which is often violated in practice due to domain shift. Domain generalization (DG) aims to achieve OOD generalization by using only source data for model learning. Over the last ten years, research in DG has made great progress, leading to a broad spectrum of methodologies, e.g., those based on domain alignment, meta-learning, data augmentation, or ensemble learning, to name a few; DG has also been studied in various application areas including computer vision, speech recognition, natural language processing, medical imaging, and reinforcement learning. In this paper, for the first time a comprehensive literature review in DG is provided to summarize the developments over the past decade. Specifically, we first cover the background by formally defining DG and relating it to other relevant fields like domain adaptation and transfer learning. Then, we conduct a thorough review into existing methods and theories. Finally, we conclude this survey with insights and discussions on future research directions.

Generalized Few-shot Semantic Segmentation (GFSS) aims to segment each image pixel into either base classes with abundant training examples or novel classes with only a handful of (e.g., 1-5) training images per class. Compared to the widely studied Few-shot Semantic Segmentation (FSS), which is limited to segmenting novel classes only, GFSS is much under-studied despite being more practical. Existing approach to GFSS is based on classifier parameter fusion whereby a newly trained novel class classifier and a pre-trained base class classifier are combined to form a new classifier. As the training data is dominated by base classes, this approach is inevitably biased towards the base classes. In this work, we propose a novel Prediction Calibration Network (PCN) to address this problem. Instead of fusing the classifier parameters, we fuse the scores produced separately by the base and novel classifiers. To ensure that the fused scores are not biased to either the base or novel classes, a new Transformer-based calibration module is introduced. It is known that the lower-level features are useful of detecting edge information in an input image than higher-level features. Thus, we build a cross-attention module that guides the classifier’s final prediction using the fused multi-level features. However, transformers are computationally demanding. Crucially, to make the proposed cross-attention module training tractable at the pixel level, this module is designed based on feature-score cross-covariance and episodically trained to be generalizable at inference time. Extensive experiments on PASCAL-$5^{i}$and COCO-$20^{i}$show that our PCN outperforms the state-the-the-art alternatives by large margins.

Source-Free Domain Adaptation (SFDA) is becoming topical to address the challenge of distribution shift between training and deployment data, while also relaxing the requirement of source data availability during target domain adaptation. In this paper, we focus on SFDA for semantic segmentation, in which pseudo labeling based target domain self-training is a common solution. However, pseudo labels generated by the source models are particularly unreliable on the target domain data due to the domain shift issue. Therefore, we propose to use Bayesian Neural Network (BNN) to improve the target self-training by better estimating and exploiting pseudo-label uncertainty. With the uncertainty estimation of BNNs, we introduce two novel self-training based components: Uncertainty-aware Online Teacher-Student Learning (UOTSL) and Uncertainty-aware FeatureMix (UFM). Extensive experiments on two popular benchmarks, GTA 5 → Cityscapes and SYNTHIA → Cityscapes, show the superiority of our proposed method with mIoU gains of 3.6% and 5.7% over the state-of-the-art respectively.

We study the practical task of fine-grained 3D-VR-sketch-based 3D shape retrieval. This task is of particular interest as 2D sketches were shown to be effective queries for 2D images. However, due to the domain gap, it remains hard to achieve strong performance in 3D shape retrieval from 2D sketches. Recent work demonstrated the advantage of 3D VR sketching on this task. In our work, we focus on the challenge caused by inherent inaccuracies in 3D VR sketches. We observe that retrieval results obtained with a triplet loss with a fixed margin value, commonly used for retrieval tasks, contain many irrelevant shapes and often just one or few with a similar structure to the query. To mitigate this problem, we for the first time draw a connection between adaptive margin values and shape similarities. In particular, we propose to use a triplet loss with an adaptive margin value driven by a ‘fitting gap’, which is the similarity of two shapes under structure-preserving deformations. We also conduct a user study which confirms that this fitting gap is indeed a suitable criterion to evaluate the structural similarity of shapes. Furthermore, we introduce a dataset of 202 VR sketches for 202 3D shapes drawn from memory rather than from observation. The code and data are available at https://github.com/Rowl1ng/Structure-Aware-VR-Sketch-Shape-Retrieval.

Graph neural networks (GNNs) have been used to tackle the few-shot learning (FSL) problem and shown great potentials under the transductive setting. However under the inductive setting, existing GNN based methods are less competitive. This is because they use an instance GNN as a label propagation/classification module, which is jointly meta-learned with a feature embedding network. This design is problematic because the classifier needs to adapt quickly to new tasks while the embedding does not. To overcome this problem, in this paper we propose a novel hybrid GNN (HGNN) model consisting of two GNNs, an instance GNN and a prototype GNN. Instead of label propagation, they act as feature embedding adaptation modules for quick adaptation of the meta-learned feature embedding to new tasks. Importantly they are designed to deal with a fundamental yet often neglected challenge in FSL, that is, with only a handful of shots per class, any few-shot classifier would be sensitive to badly sampled shots which are either outliers or can cause inter-class distribution overlapping. Extensive experiments show that our HGNN obtains new state-of-the-art on three FSL benchmarks. The code and models are available at https://github.com/TianyuanYu/HGNN.

The human visual system is remarkable in learning new visual concepts from just a few examples. This is precisely the goal behind few-shot class incremental learning (FS-CIL), where the emphasis is additionally placed on ensuring the model does not suffer from “forgetting”. In this paper, we push the boundary further for FSCIL by addressing two key questions that bottleneck its ubiquitous application (i) can the model learn from diverse modalities other than just photo (as humans do), and (ii) what if photos are not readily accessible (due to ethical and privacy constraints). Our key innovation lies in advocating the use of sketches as a new modality for class support. The product is a “Doodle It Yourself” (DIY) FSCIL framework where the users can freely sketch a few examples of a novel class for the model to learn to recognise photos of that class. For that, we present a framework that infuses (i) gradient consensus for domain invariant learning, (ii) knowledge distillation for preserving old class information, and (iii) graph attention networks for message passing between old and novel classes. We experimentally show that sketches are better class support than text in the context of FSCIL, echoing findings elsewhere in the sketching literature.

Sketching enables many exciting applications, notably, image retrieval. The fear-to-sketch problem (i.e., “I can't sketch”) has however proven to be fatal for its widespread adoption. This paper tackles this “fear” head on, and for the first time, proposes an auxiliary module for existing retrieval models that predominantly lets the users sketch without having to worry. We first conducted a pilot study that revealed the secret lies in the existence of noisy strokes, but not so much of the “I can't sketch”. We consequently design a stroke subset selector that detects noisy strokes, leaving only those which make a positive contribution towards successful retrieval. Our Reinforcement Learning based formulation quantifies the importance of each stroke present in a given subset, based on the extent to which that stroke contributes to retrieval. When combined with pre-trained retrieval models as a pre-processing module, we achieve a significant gain of 8%-10% over standard baselines and in turn report new state-of-the-art performance. Last but not least, we demonstrate the selector once trained, can also be used in a plug-and-play manner to empower various sketch applications in ways that were not previously possible.

We scrutinise an important observation plaguing scene-level sketch research - that a significant portion of scene sketches are “partial”. A quick pilot study reveals: (i) a scene sketch does not necessarily contain all objects in the corresponding photo, due to the subjective holistic interpretation of scenes, (ii) there exists significant empty (white) regions as a result of object-level abstraction, and as a result, (iii) existing scene-level fine-grained sketch-based image retrieval methods collapse as scene sketches become more partial. To solve this “partial” problem, we advocate for a simple set-based approach using optimal transport (OT) to model cross-modal region associativity in a partially-aware fashion. Importantly, we improve upon OT to further account for holistic partialness by comparing intra-modal adjacency matrices. Our proposed method is not only robust to partial scene-sketches but also yields state-of-the-art performance on existing datasets.

Image-based virtual try-on aims to fit an in-shop garment into a clothed person image. To achieve this, a key step is garment warping which spatially aligns the target garment with the corresponding body parts in the person image. Prior methods typically adopt a local appearance flow estimation model. They are thus intrinsically susceptible to difficult body poses/occlusions and large mis-alignments between person and garment images (see Fig. 1). To overcome this limitation, a novel global appearance flow estimation model is proposed in this work. For the first time, a StyleGAN based architecture is adopted for appearance flow estimation. This enables us to take advantage of a global style vector to encode a whole-image context to cope with the aforementioned challenges. To guide the StyleGAN flow generator to pay more attention to local garment deformation, a flow refinement module is introduced to add local context. Experiment results on a popular virtual tryon benchmark show that our method achieves new state-of-the-art performance. It is particularly effective in a ‘in-the-wild’ application scenario where the reference image is full-body resulting in a large mis-alignment with the garment image (Fig. 1 Top). Code is available at: https://github.com/SenHe/Flow-Style-VTON.

Zero-shot sketch-based image retrieval typically asks for a trained model to be applied as is to unseen categories. In this paper, we question to argue that this setup by definition is not compatible with the inherent abstract and subjective nature of sketches – the model might transfer well to new categories, but will not understand sketches existing in different test-time distribution as a result. We thus extend ZS-SBIR asking it to transfer to both categories and sketch distributions. Our key contribution is a test-time training paradigm that can adapt using just one sketch. Since there is no paired photo, we make use of a sketch raster-vector reconstruction module as a self-supervised auxiliary task. To maintain the fidelity of the trained cross-modal joint embedding during test-time update, we design a novel meta-learning based training paradigm to learn a separation between model updates incurred by this auxiliary task from those off the primary objective of discriminative learning. Extensive experiments show our model to outperform state-of-the-arts, thanks to the proposed test-time adaption that not only transfers to new categories but also accommodates to new sketching styles.

We advance sketch research to scenes with the first dataset of freehand scene sketches, FS-COCO. With practical applications in mind, we collect sketches that convey scene content well but can be sketched within a few minutes by a person with any sketching skills. Our dataset comprises 10, 000 freehand scene vector sketches with per point space-time information by 100 non-expert individuals, offering both object- and scene-level abstraction. Each sketch is augmented with its text description. Using our dataset, we study for the first time the problem of fine-grained image retrieval from freehand scene sketches and sketch captions. We draw insights on: (i) Scene salience encoded in sketches using the strokes temporal order; (ii) Performance comparison of image retrieval from a scene sketch and an image caption; (iii) Complementarity of information in sketches and image captions, as well as the potential benefit of combining the two modalities. In addition, we extend a popular vector sketch LSTM-based encoder to handle sketches with larger complexity than was supported by previous work. Namely, we propose a hierarchical sketch decoder, which we leverage at a sketch-specific “pretext” task. Our dataset enables for the first time research on freehand scene sketch understanding and its practical applications. We release the dataset under CC BY-NC 4.0 license: FS-COCO dataset ( https://github.com/pinakinathc/fscoco ).

Multimodal tasks in the fashion domain have significant potential for e-commerce, but involve challenging vision-and-language learning problems-e.g., retrieving a fashion item given a reference image plus text feedback from a user.Prior works on multimodal fashion tasks have either been limited by the data in individual benchmarks, or have leveraged generic vision-and-language pre-training but have not taken advantage of the characteristics of fashion data.Additionally, these works have mainly been restricted to multimodal understanding tasks.To address these gaps, we make two key contributions.First, we propose a novel fashion-specific pre-training framework based on weakly-supervised triplets constructed from fashion image-text pairs.We show the tripletbased tasks are an effective addition to standard multimodal pre-training tasks.Second, we propose a flexible decoder-based model architecture capable of both fashion retrieval and captioning tasks.Together, our model design and pre-training approach are competitive on a diverse set of fashion tasks, including crossmodal retrieval, image retrieval with text feedback, image captioning, relative image captioning, and multimodal categorization.

Person re-identification (Re-ID) aims at retrieving a person of interest across multiple non-overlapping cameras. With the advancement of deep neural networks and increasing demand of intelligent video surveillance, it has gained significantly increased interest in the computer vision community. By dissecting the involved components in developing a person Re-ID system, we categorize it into the closed-world and open-world settings. The widely studied closed-world setting is usually applied under various research-oriented assumptions, and has achieved inspiring success using deep learning techniques on a number of datasets. We first conduct a comprehensive overview with in-depth analysis for closed-world person Re-ID from three different perspectives, including deep feature representation learning, deep metric learning and ranking optimization. With the performance saturation under closed-world setting, the research focus for person Re-ID has recently shifted to the open-world setting, facing more challenging issues. This setting is closer to practical applications under specific scenarios. We summarize the open-world Re-ID in terms of five different aspects. By analyzing the advantages of existing methods, we design a powerful AGW baseline, achieving state-of-the-art or at least comparable performance on twelve datasets for four different Re-ID tasks. Meanwhile, we introduce a new evaluation metric (mINP) for person Re-ID, indicating the cost for finding all the correct matches, which provides an additional criteria to evaluate the Re-ID system for real applications. Finally, some important yet under-investigated open issues are discussed.

An effective person re-identification (re-ID) model should learn feature representations that are both discriminative, for distinguishing similar-looking people, and generalisable, for deployment across datasets without any adaptation. In this paper, we develop novel CNN architectures to address both challenges. First, we present a re-ID CNN termed omni-scale network (OSNet) to learn features that not only capture different spatial scales but also encapsulate a synergistic combination of multiple scales, namely omni-scale features. The basic building block consists of multiple convolutional streams, each detecting features at a certain scale. For omni-scale feature learning, a unified aggregation gate is introduced to dynamically fuse multi-scale features with channel-wise weights. OSNet is lightweight as its building blocks comprise factorised convolutions. Second, to improve generalisable feature learning, we introduce instance normalisation (IN) layers into OSNet to cope with cross-dataset discrepancies. Further, to determine the optimal placements of these IN layers in the architecture, we formulate an efficient differentiable architecture search algorithm. Extensive experiments show that, in the conventional same-dataset setting, OSNet achieves state-of-the-art performance, despite being much smaller than existing re-ID models. In the more challenging yet practical cross-dataset setting, OSNet beats most recent unsupervised domain adaptation methods without using any target data.

Most existing studies on unsupervised domain adaptation (UDA) assume that each domain's training samples come with domain labels (e.g., painting, photo). Samples from each domain are assumed to follow the same distribution and the domain labels are exploited to learn domain-invariant features via feature alignment. However, such an assumption often does not hold true-there often exist numerous finer-grained domains (e.g., dozens of modern painting styles have been developed, each differing dramatically from those of the classic styles). Therefore, forcing feature distribution alignment across each artificially-defined and coarse-grained domain can be ineffective. In this paper, we address both single-source and multi-source UDA from a completely different perspective, which is to view each instance as a fine domain. Feature alignment across domains is thus redundant. Instead, we propose to perform dynamic instance domain adaptation (DIDA). Concretely, a dynamic neural network with adaptive convolutional kernels is developed to generate instance-adaptive residuals to adapt domain-agnostic deep features to each individual instance. This enables a shared classifier to be applied to both source and target domain data without relying on any domain annotation. Further, instead of imposing intricate feature alignment losses, we adopt a simple semi-supervised learning paradigm using only a cross-entropy loss for both labeled source and pseudo labeled target data. Our model, dubbed DIDA-Net, achieves state-of-the-art performance on several commonly used single-source and multi-source UDA datasets including Digits, Office-Home, DomainNet, Digit-Five, and PACS.

We present the first one-shot personalized sketch segmentation method. We aim to segment all sketches belonging to the same category provisioned with a single sketch with a given part annotation while (i) preserving the parts semantics embedded in the exemplar, and (ii) being robust to input style and abstraction. We refer to this scenario as personalized. With that, we importantly enable a much-desired personalization capability for downstream fine-grained sketch analysis tasks. To train a robust segmentation module, we deform the exemplar sketch to each of the available sketches of the same category. Our method generalizes to sketches not observed during training. Our central contribution is a sketch-specific hierarchical deformation network. Given a multi-level sketch-strokes encoding obtained via a graph convolutional network, our method estimates rigid-body transformation from the target to the exemplar, on the upper level. Finer deformation from the exemplar to the globally warped target sketch is further obtained through stroke-wise deformations, on the lower-level. Both levels of deformation are guided by mean squared distances between the keypoints learned without supervision, ensuring that the stroke semantics are preserved. We evaluate our method against the state-of-the-art segmentation and perceptual grouping baselines re-purposed for the one-shot setting and against two few-shot 3D shape segmentation methods. We show that our method outperforms all the alternatives by more than 10% on average. Ablation studies further demonstrate that our method is robust to personalization: changes in input part semantics and style differences.

We present the first fine-grained dataset of 1,497 3D VR sketch and 3D shape pairs of a chair category with large shapes diversity. Our dataset supports the recent trend in the sketch community on fine-grained data analysis, and extends it to an actively developing 3D domain. We argue for the most convenient sketching scenario where the sketch consists of sparse lines and does not require any sketching skills, prior training or time-consuming accurate drawing. We then, for the first time, study the scenario of fine-grained 3D VR sketch to 3D shape retrieval, as a novel VR sketching application and a proving ground to drive out generic insights to inform future research. By experimenting with carefully selected combinations of design factors on this new problem, we draw important conclusions to help follow-on work. We hope our dataset will enable other novel applications, especially those that require a fine-grained angle such as fine-grained 3D shape reconstruction. The dataset is available at tinyurl.com/ VRSketch3DV21.

Self-Supervised monocular visual odometry (VO) is often cast into a view synthesis problem based on depth and camera pose estimation. One of the key challenges is to accurately and robustly estimate depth with occlusions and moving objects in the scene. Existing methods simply detect and mask out regions of occlusions locally by several convolutional layers, and then perform only partial view synthesis in the rest of the image. However, occlusion and moving object detection is an unsolved problem itself which requires global layout information. Inaccurate detection inevitably results in incorrect depth as well as pose estimation. In this work, instead of locally detecting and masking out occlusions and moving objects, we propose to alleviate their negative effects on monocular VO implicitly but more effectively from two global perspectives. First, a multi-scale non-local attention module, consisting of both intra-stage augmented attention and cascaded across-stage attention, is proposed for robust depth estimation given occlusions, alleviating the impacts of occlusions via global attention modeling. Second, adversarial learning is introduced in view synthesis for monocular VO. Unlike existing methods that use pixel-level losses on the quality of synthesized views, we enforce the synthetic view to be indistinguishable from the real one at the scene-level. Such a global constraint again helps cope with occluded and moving regions. Extensive experiments on the KITTI dataset show that our approach achieves new state-of-the-art in both pose estimation and depth recovery.

Stochastic neural networks (SNNs) are currently topical, with several paradigms being actively investigated including dropout, Bayesian neural networks, variational information bottleneck (VIB) and noise regularized learning. These neural network variants impact several major considerations, including generalization, network compression, robustness against adversarial attack and label noise, and model calibration. However, many existing networks are complicated and expensive to train, and/or only address one or two of these practical considerations. In this paper we propose a simple and effective stochastic neural network (SE-SNN) architecture for discriminative learning by directly modeling activation uncertainty and encouraging high activation variability. Compared to existing SNNs, our SE-SNN is simpler to implement and faster to train, and produces state of the art results on network compression by pruning, adversarial defense, learning with label noise, and model calibration.

Analysis of human sketches in deep learning has advanced immensely through the use of waypoint-sequences rather than raster-graphic representations. We further aim to model sketches as a sequence of low-dimensional parametric curves. To this end, we propose an inverse graphics framework capable of approximating a raster or waypoint based stroke encoded as a point-cloud with a variable-degree Bézier curve. Building on this module, we present Cloud2Curve, a generative model for scalable high-resolution vector sketches that can be trained end-to-end using point-cloud data alone. As a consequence, our model is also capable of deterministic vectorization which can map novel raster or waypoint based sketches to their corresponding high-resolution scalable Bézier equivalent. We evaluate the generation and vectorization capabilities of our model on Quick, Draw! and K-MNIST datasets.

A fundamental challenge faced by existing Fine-Grained Sketch-Based Image Retrieval (FG-SBIR) models is the data scarcity – model performances are largely bottlenecked by the lack of sketch-photo pairs. Whilst the number of photos can be easily scaled, each corresponding sketch still needs to be individually produced. In this paper, we aim to mitigate such an upper-bound on sketch data, and study whether unlabelled photos alone (of which they are many) can be cultivated for performance gain. In particular, we introduce a novel semi-supervised framework for cross-modal retrieval that can additionally leverage large-scale unlabelled photos to account for data scarcity. At the center of our semi-supervision design is a sequential photo-to-sketch generation model that aims to generate paired sketches for unlabelled photos. Importantly, we further introduce a discriminator-guided mechanism to guide against unfaithful generation, together with a distillation loss-based regularizer to provide tolerance against noisy training samples. Last but not least, we treat generation and retrieval as two conjugate problems, where a joint learning procedure is devised for each module to mutually benefit from each other. Extensive experiments show that our semi-supervised model yields a significant performance boost over the state-of-the-art supervised alternatives, as well as existing methods that can exploit unlabelled photos for FG-SBIR.

Self-supervised learning has gained prominence due to its efficacy at learning powerful representations from unlabelled data that achieve excellent performance on many challenging downstream tasks. However, supervision-free pre-text tasks are challenging to design and usually modality specific. Although there is a rich literature of self-supervised methods for either spatial (such as images) or temporal data (sound or text) modalities, a common pretext task that benefits both modalities is largely missing. In this paper, we are interested in defining a self-supervised pre-text task for sketches and handwriting data. This data is uniquely characterised by its existence in dual modalities of rasterized images and vector coordinate sequences. We address and exploit this dual representation by proposing two novel cross-modal translation pre-text tasks for self-supervised feature learning: Vectorization and Rasterization. Vectorization learns to map image space to vector coordinates and rasterization maps vector coordinates to image space. We show that our learned encoder modules benefit both raster-based and vector-based downstream approaches to analysing hand-drawn data. Empirical evidence shows that our novel pre-text tasks surpass existing single and multi-modal self-supervision methods.

A layout to image (L2I) generation model aims to generate a complicated image containing multiple objects (things) against natural background (stuff), conditioned on a given layout. Built upon the recent advances in generative adversarial networks (GANs), existing L2I models have made great progress. However, a close inspection of their generated images reveals two major limitations: (1) the object-to-object as well as object-to-stuff relations are often broken and (2) each object’s appearance is typically distorted lacking the key defining characteristics associated with the object class. We argue that these are caused by the lack of context-aware object and stuff feature encoding in their generators, and location-sensitive appearance representation in their discriminators. To address these limitations, two new modules are proposed in this work. First, a context-aware feature transformation module is introduced in the generator to ensure that the generated feature encoding of either object or stuff is aware of other coexisting objects/stuff in the scene. Second, instead of feeding location-insensitive image features to the discriminator, we use the Gram matrix computed from the feature maps of the generated object images to preserve location-sensitive information, resulting in much enhanced object appearance. Extensive experiments show that the proposed method achieves state-of-the-art performance on the COCO-Thing-Stuff and Visual Genome benchmarks. Code available at: https://github.com/wtliao/layout2img.

Sketch-based image retrieval (SBIR) is a cross-modal matching problem which is typically solved by learning a joint embedding space where the semantic content shared between photo and sketch modalities are preserved. However, a fundamental challenge in SBIR has been largely ignored so far, that is, sketches are drawn by humans and considerable style variations exist amongst different users. An effective SBIR model needs to explicitly account for this style diversity, crucially, to generalise to unseen user styles. To this end, a novel style-agnostic SBIR model is proposed. Different from existing models, a cross-modal variational autoencoder (VAE) is employed to explicitly disentangle each sketch into a semantic content part shared with the corresponding photo, and a style part unique to the sketcher. Importantly, to make our model dynamically adaptable to any unseen user styles, we propose to meta-train our cross-modal VAE by adding two style-adaptive components: a set of feature transformation layers to its encoder and a regulariser to the disentangled semantic content latent code. With this meta-learning framework, our model can not only disentangle the cross-modal shared semantic content for SBIR, but can adapt the disentanglement to any unseen user style as well, making the SBIR model truly style-agnostic. Extensive experiments show that our style-agnostic model yields state-of-the-art performance for both category-level and instance-level SBIR.

A deep facial attribute editing model strives to meet two requirements: (1) attribute correctness – the target attribute should correctly appear on the edited face image; (2) irrelevance preservation – any irrelevant information (e.g., identity) should not be changed after editing. Meeting both requirements challenges the state-of-the-art works which resort to either spatial attention or latent space factorization. Specifically, the former assume that each attribute has well-defined local support regions; they are often more effective for editing a local attribute than a global one. The latter factorize the latent space of a fixed pretrained GAN into different attribute-relevant parts, but they cannot be trained end-to-end with the GAN, leading to sub-optimal solutions. To overcome these limitations, we propose a novel latent space factorization model, called L2M-GAN, which is learned end-to-end and effective for editing both local and global attributes. The key novel components are: (1) A latent space vector of the GAN is factorized into an attribute-relevant and irrelevant codes with an orthogonality constraint imposed to ensure disentanglement. (2) An attribute-relevant code transformer is learned to manipulate the attribute value; crucially, the transformed code are subject to the same orthogonality constraint. By forcing both the original attribute-relevant latent code and the edited code to be disentangled from any attribute-irrelevant code, our model strikes the perfect balance between attribute correctness and irrelevance preservation. Extensive experiments on CelebA-HQ show that our L2M-GAN achieves significant improvements over the state-of-the-arts.

Most recent semantic segmentation methods adopt a fully-convolutional network (FCN) with an encoder-decoder architecture. The encoder progressively reduces the spatial resolution and learns more abstract/semantic visual concepts with larger receptive fields. Since context modeling is critical for segmentation, the latest efforts have been focused on increasing the receptive field, through either dilated/atrous convolutions or inserting attention modules. However, the encoder-decoder based FCN architecture remains unchanged. In this paper, we aim to provide an alternative perspective by treating semantic segmentation as a sequence-to-sequence prediction task. Specifically, we deploy a pure transformer (i.e., without convolution and resolution reduction) to encode an image as a sequence of patches. With the global context modeled in every layer of the transformer, this encoder can be combined with a simple decoder to provide a powerful segmentation model, termed SEgmentation TRansformer (SETR). Extensive experiments show that SETR achieves new state of the art on ADE20K (50.28% mIoU), Pascal Context (55.83% mIoU) and competitive results on Cityscapes. Particularly, we achieve the first position in the highly competitive ADE20K test server leaderboard on the day of submission.

The goal of few-shot learning (FSL) is to recognize a set of novel classes with only few labeled samples by exploiting a large set of abundant base class samples. Adopting a meta-learning framework, most recent FSL methods meta-learn a deep feature embedding network, and during inference classify novel class samples using nearest neighbor in the learned high-dimensional embedding space. This means that these methods are prone to the hubness problem, that is, a certain class prototype becomes the nearest neighbor of many test instances regardless which classes they belong to. However, this problem is largely ignored in existing FSL studies. In this work, for the first time we show that many FSL methods indeed suffer from the hubness problem. To mitigate its negative effects, we further propose to employ z-score feature normalization, a simple yet effective trans-formation, during meta-training. A theoretical analysis is provided on why it helps. Extensive experiments are then conducted to show that with z-score normalization, the performance of many recent FSL methods can be boosted, resulting in new state-of-the-art on three benchmarks.

A lifespan face synthesis (LFS) model aims to generate a set of photo-realistic face images of a person’s whole life, given only one snapshot as reference. The generated face image given a target age code is expected to be age-sensitive reflected by bio-plausible transformations of shape and texture, while being identity preserving. This is extremely challenging because the shape and texture characteristics of a face undergo separate and highly nonlinear transformations w.r.t. age. Most recent LFS models are based on generative adversarial networks (GANs) whereby age code conditional transformations are applied to a latent face representation. They benefit greatly from the recent advancements of GANs. However, without explicitly disentangling their latent representations into the texture, shape and identity factors, they are fundamentally limited in modeling the nonlinear age-related transformation on texture and shape whilst preserving identity. In this work, a novel LFS model is proposed to disentangle the key face characteristics including shape, texture and identity so that the unique shape and texture age transformations can be modeled effectively. This is achieved by extracting shape, texture and identity features separately from an encoder. Critically, two transformation modules, one conditional convolution based and the other channel attention based, are designed for modeling the nonlinear shape and texture feature transformations respectively. This is to accommodate their rather distinct aging processes and ensure that our synthesized images are both age-sensitive and identity preserving. Extensive experiments show that our LFS model is clearly superior to the state-of-the-art alternatives. Codes and demo are available on our project website: https://senhe.github.io/projects/iccv_2021_lifespan_face.

A few-shot semantic segmentation model is typically composed of a CNN encoder, a CNN decoder and a simple classifier (separating foreground and background pixels). Most existing methods meta-learn all three model components for fast adaptation to a new class. However, given that as few as a single support set image is available, effective model adaption of all three components to the new class is extremely challenging. In this work we propose to simplify the meta-learning task by focusing solely on the simplest component – the classifier, whilst leaving the en-coder and decoder to pre-training. We hypothesize that if we pretrain an off-the-shelf segmentation model over a set of diverse training classes with sufficient annotations, the encoder and decoder can capture rich discriminative features applicable for any unseen classes, rendering the sub-sequent meta-learning stage unnecessary. For the classifier meta-learning, we introduce a Classifier Weight Transformer (CWT) designed to dynamically adapt the support-set trained classifier’s weights to each query image in an inductive way. Extensive experiments on two standard bench-marks show that despite its simplicity, our method outperforms the state-of-the-art alternatives, often by a large margin. Code is available on https://github.com/zhiheLu/CWT-for-FSS.

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 proposes a novel pretext task for self-supervised video representation learning by exploiting spatiotemporal continuity in videos. It is motivated by the fact that videos are spatiotemporal by nature and a representation learned by detecting spatiotemporal continuity/discontinuity is thus beneficial for downstream video content analysis tasks. A natural choice of such a pretext task is to construct spatiotemporal (3D) jigsaw puzzles and learn to solve them. However, as we demonstrate in the experiments, this task turns out to be intractable. We thus propose Constrained Spatiotemporal Jigsaw (CSJ) whereby the 3D jigsaws are formed in a constrained manner to ensure that large continuous spatiotemporal cuboids exist. This provides sufficient cues for the model to reason about the continuity. Instead of solving them directly, which could still be extremely hard, we carefully design four surrogate tasks that are more solvable. The four tasks aim to learn representations sensitive to spatiotemporal continuity at both the local and global levels. Extensive experiments show that our CSJ achieves state-of-the-art on various benchmarks.

A deep hashing model typically has two main learning objectives: to make the learned binary hash codes discriminative and to minimize a quantization error. With further constraints such as bit balance and code orthogonality, it is not uncommon for existing models to employ a large number (>4) of losses. This leads to difficulties in model training and subsequently impedes their effectiveness. In this work, we propose a novel deep hashing model with only $\textit{a single learning objective}$. Specifically, we show that maximizing the cosine similarity between the continuous codes and their corresponding $\textit{binary orthogonal codes}$ can ensure both hash code discriminativeness and quantization error minimization. Further, with this learning objective, code balancing can be achieved by simply using a Batch Normalization (BN) layer and multi-label classification is also straightforward with label smoothing. The result is a one-loss deep hashing model that removes all the hassles of tuning the weights of various losses. Importantly, extensive experiments show that our model is highly effective, outperforming the state-of-the-art multi-loss hashing models on three large-scale instance retrieval benchmarks, often by significant margins.

Vision transformers (ViTs) have pushed the state-of-the-art for various visual recognition tasks by patch-wise image tokenization followed by self-attention. However, the employment of self-attention modules results in a quadratic complexity in both computation and memory usage. Various attempts on approximating the self-attention computation with linear complexity have been made in Natural Language Processing. However, an in-depth analysis in this work shows that they are either theoretically flawed or empirically ineffective for visual recognition. We further identify that their limitations are rooted in keeping the softmax self-attention during approximations. Specifically, conventional self-attention is computed by normalizing the scaled dot-product between token feature vectors. Keeping this softmax operation challenges any subsequent linearization efforts. Based on this insight, for the first time, a softmax-free transformer or SOFT is proposed. To remove softmax in self-attention, Gaussian kernel function is used to replace the dot-product similarity without further normalization. This enables a full self-attention matrix to be approximated via a low-rank matrix decomposition. The robustness of the approximation is achieved by calculating its Moore-Penrose inverse using a Newton-Raphson method. Extensive experiments on ImageNet show that our SOFT significantly improves the computational efficiency of existing ViT variants. Crucially, with a linear complexity, much longer token sequences are permitted in SOFT, resulting in superior trade-off between accuracy and complexity.

Most recent few-shot learning (FSL) methods are based on meta-learning with episodic training. In each meta-training episode, a discriminative feature embedding and/or classifier are first constructed from a support set in an inner loop, and then evaluated in an outer loop using a query set for model updating. This query set sample centered learning objective is however intrinsically limited in addressing the lack of training data problem in the support set. In this paper, a novel contrastive prototype learning with augmented embeddings (CPLAE) model is proposed to overcome this limitation. First, data augmentations are introduced to both the support and query sets with each sample now being represented as an augmented embedding (AE) composed of concatenated embeddings of both the original and augmented versions. Second, a novel support set class prototype centered contrastive loss is proposed for contrastive prototype learning (CPL). With a class prototype as an anchor, CPL aims to pull the query samples of the same class closer and those of different classes further away. This support set sample centered loss is highly complementary to the existing query centered loss, fully exploiting the limited training data in each episode. Extensive experiments on several benchmarks demonstrate that our proposed CPLAE achieves new state-of-the-art.

Detecting subtle intention such as deception and subtext of a person in a long dialog video, or implicit intention detection (IID), is a challenging problem. The transcript (textual cues) often reveals little, so audio-visual cues including voice tone as well as facial and body behaviour are the main focuses for automated IID. Contextual cues are also crucial, since a person's implicit intentions are often correlated and context-dependent when the person moves from one question-answer pair to the next. However, no such dataset exists which contains fine-grained question-answer pair (video segment) level annotation. The first contribution of this work is thus a new benchmark dataset, called Vid2Int-Deception to fill this gap. A novel multigrain representation model is also proposed to capture the subtle movement changes of eyes, face, and body (relevant for inferring intention) from a long dialog video. Moreover, to model the temporal correlation between the implicit intentions across video segments, we propose a Video-to-Intention network (Vid2Int) based on attentive recurrent neural network (RNN). Extensive experiments show that our model achieves state-of-the-art results.

Existing few-shot learning (FSL) methods assume that there exist sufficient training samples from source classes for knowledge transfer to target classes with few training samples. However, this assumption is often invalid, especially when it comes to fine-grained recognition. In this work, we define a new FSL setting termed scarce-source few-shot learning (SSFSL), under which both the source and target classes have limited training samples. To overcome the source class data scarcity problem, a natural option is to crawl images from the web with class names as search keywords. However, the crawled images are inevitably corrupted by large amount of noise (irrelevant images) and thus may harm the performance. To address this problem, we propose a graph convolutional network (GCN)-based label denoising (LDN) method to remove the irrelevant images. Further, with the cleaned web images as well as the original clean training images, we propose a GCN-based FSL method. For both the LDN and FSL tasks, a novel adaptive aggregation GCN (AdarGCN) model is proposed, which differs from existing GCN models in that adaptive aggregation is performed based on a multi-head multi-level aggregation module. With AdarGCN, how much and how far information carried by each graph node is propagated in the graph structure can be determined automatically, therefore alleviating the effects of both noisy and outlying training samples. Extensive experiments demonstrate the superior performance of our AdarGCN under both the new SSFSL and the conventional FSL settings.

Existing few-shot learning (FSL) methods make the implicit assumption that the few target class samples are from the same domain as the source class samples. However, in practice, this assumption is often invalid -the target classes could come from a different domain. This poses an additional challenge of domain adaptation (DA) with few training samples. In this paper, the problem of domain-adaptive few-shot learning (DA-FSL) is tackled, which is expected to have wide use in real-world scenarios and requires solving FSL and DA in a unified framework. To this end, we propose a novel domain-adversarial prototypical network (DAPN) model. It is designed to address a specific challenge in DA-FSL: the DA objective means that the source and target data distributions need to be aligned, typically through a shared domain-adaptive feature embedding space; but the FSL objective dictates that the target domain per class distribution must be different from that of any source domain class, meaning aligning the distributions across domains may harm the FSL performance. How to achieve global domain distribution alignment whilst maintaining source/target per-class discriminativeness thus becomes the key. Our solution is to explicitly enhance the source/target per-class separation before domain-adaptive feature embedding learning, to alleviate the negative effect of domain alignment on FSL. Extensive experiments show that our DAPN outperforms the state-of-the-arts. The code is available at https://github.com/dingmyu/DAPN.

Zero-shot learning (ZSL) is made possible by learning a projection function between a feature space and a semantic space (e.g., an attribute space). Key to ZSL is thus to learn a projection that is robust against the often large domain gap between the seen and unseen class domains. In this work, this is achieved by unseen class data synthesis and robust projection function learning. Specifically, a novel semantic data synthesis strategy is proposed, by which semantic class prototypes (e.g., attribute vectors) are used to simply perturb seen class data for generating unseen class ones. As in any data synthesis/hallucination approach, there are ambiguities and uncertainties on how well the synthesised data can capture the targeted unseen class data distribution. To cope with this, the second contribution of this work is a novel projection learning model termed competitive bidirectional projection learning (BPL) designed to best utilise the ambiguous synthesised data. Specifically, we assume that each synthesised data point can belong to any unseen class; and the most likely two class candidates are exploited to learn a robust projection function in a competitive fashion. As a third contribution, we show that the proposed ZSL model can be easily extended to few-shot learning (FSL) by again exploiting semantic (class prototype guided) feature synthesis and competitive BPL. Extensive experiments show that our model achieves the state-of-the-art results on both problems.

In this paper, we focus on learning semantic representations for large-scale highly abstract sketches that were produced by the practical sketch-based application rather than the excessively well dawn sketches obtained by crowd-sourcing. We propose a dual-branch CNN-RNN network architecture to represent sketches, which simultaneously encodes both the static and temporal patterns of sketch strokes. Based on this architecture, we further explore learning the sketch-oriented semantic representations in two practical settings, i.e., hashing retrieval and zero-shot recognition on million-scale highly abstract sketches produced by practical online interactions. Specifically, we use our dual-branch architecture as a universal representation framework to design two sketch-specific deep models: (i) We propose a deep hashing model for sketch retrieval, where a novel hashing loss is specifically designed to further accommodate both the abstract and messy traits of sketches. (ii) We propose a deep embedding model for sketch zero-shot recognition, via collecting a large-scale edge-map dataset and proposing to extract a set of semantic vectors from edge-maps as the semantic knowledge for sketch zero-shot domain alignment. Both deep models are evaluated by comprehensive experiments on million-scale abstract sketches produced by a global online game QuickDraw and outperform state-of-the-art competitors.

Existing sketch-analysis work studies sketches depicting static objects or scenes. In this work, we propose a novel cross-modal retrieval problem of fine-grained instance-level sketch-based video retrieval (FG-SBVR), where a sketch sequence is used as a query to retrieve a specific target video instance. Compared with sketch-based still image retrieval, and coarse-grained category-level video retrieval, this is more challenging as both visual appearance and motion need to be simultaneously matched at a fine-grained level. We contribute the first FG-SBVR dataset with rich annotations. We then introduce a novel multi-stream multi-modality deep network to perform FG-SBVR under both strong and weakly supervised settings. The key component of the network is a relation module, designed to prevent model overfitting given scarce training data. We show that this model significantly outperforms a number of existing state-of-the-art models designed for video analysis.

Modelling long-range contextual relationships is critical for pixel-wise prediction tasks such as semantic segmentation. However, convolutional neural networks (CNNs) are inherently limited to model such dependencies due to the naive structure in its building modules (e.g., local convolution kernel). While recent global aggregation methods are beneficial for long-range structure information modelling, they would oversmooth and bring noise to the regions contain fine details (e.g., boundaries and small objects), which are very much cared in the semantic segmentation task. To alleviate this problem, we propose to explore the local context for making the aggregated long-range relationship being distributed more accurately in local regions. In particular, we design a novel local distribution module which models the affinity map between global and local relationship for each pixel adaptively. Integrating existing global aggregation modules, we show that our approach can be modularized as an end-to-end trainable block and easily plugged into existing semantic segmentation networks, giving rise to the GALD networks. Despite its simplicity and versatility, our approach allows us to build new state of the art on major semantic segmentation benchmarks including Cityscapes, ADE20K, Pascal Context, Camvid and COCO-stuff. Code and trained models are released at https://github.com/lxtGH/GALD-DGCNet to foster further research.

In this paper we study, for the first time, the problem of fine-grained sketch-based 3D shape retrieval. We advocate the use of sketches as a fine-grained input modality to retrieve 3D shapes at instance-level - e.g., given a sketch of a chair, we set out to retrieve a specific chair from a gallery of all chairs. Fine-grained sketch-based 3D shape retrieval (FG-SBSR) has not been possible till now due to a lack of datasets that exhibit one-to-one sketch-3D correspondences. The first key contribution of this paper is two new datasets, consisting a total of 4,680 sketch-3D pairings from two object categories. Even with the datasets, FG-SBSR is still highly challenging because (i) the inherent domain gap between 2D sketch and 3D shape is large, and (ii) retrieval needs to be conducted at the instance level instead of the coarse category level matching as in traditional SBSR. Thus, the second contribution of the paper is the first cross-modal deep embedding model for FG-SBSR, which specifically tackles the unique challenges presented by this new problem. Core to the deep embedding model is a novel cross-modal view attention module which automatically computes the optimal combination of 2D projections of a 3D shape given a query sketch.

The problem of generalizing deep neural networks from multiple source domains to a target one is studied under two settings: When unlabeled target data is available, it is a multi-source unsupervised domain adaptation (UDA) problem, otherwise a domain generalization (DG) problem. We propose a unified framework termed domain adaptive ensemble learning (DAEL) to address both problems. A DAEL model is composed of a CNN feature extractor shared across domains and multiple classifier heads each trained to specialize in a particular source domain. Each such classifier is an expert to its own domain but a non-expert to others. DAEL aims to learn these experts collaboratively so that when forming an ensemble, they can leverage complementary information from each other to be more effective for an unseen target domain. To this end, each source domain is used in turn as a pseudo-target-domain with its own expert providing supervisory signal to the ensemble of non-experts learned from the other sources. To deal with unlabeled target data under the UDA setting where real expert does not exist, DAEL uses pseudo labels to supervise the ensemble learning. Extensive experiments on three multi-source UDA datasets and two DG datasets show that DAEL improves the state of the art on both problems, often by significant margins.

Growing free online 3D shapes collections dictated research on 3D retrieval. Active debate has however been had on (i) what the best input modality is to trigger retrieval, and (ii) the ultimate usage scenario for such retrieval. In this paper, we offer a different perspective towards answering these questions - we study the use of 3D sketches as an input modality and advocate a VR-scenario where retrieval is conducted. Thus, the ultimate vision is that users can freely retrieve a 3D model by air-doodling in a VR environment. As a first stab at this new 3D VR-sketch to 3D shape retrieval problem, we make four contributions. First, we code a VR utility to collect 3D VR-sketches and conduct retrieval. Second, we collect the first set of 167 3D VRsketches on two shape categories from ModelNet. Third, we propose a novel approach to generate a synthetic dataset of human-like 3D sketches of different abstract levels to train deep networks. At last, we compare the common multi-view and volumetric approaches: We show that, in contrast to 3D shape to 3D shape retrieval, volumetric point-based approaches exhibit superior performance on 3D sketch to 3D shape retrieval due to the sparse and abstract nature of 3D VR-sketches. We believe these contributions will collectively serve as enablers for future attempts at this problem. The VR interface, code and datasets are available at https://tinyurl.com/3DSketch3DV.

Machine learning models typically suffer from the domain shift problem when trained on a source dataset and evaluated on a target dataset of different distribution. To overcome this problem, domain generalisation (DG) methods aim to leverage data from multiple source domains so that a trained model can generalise to unseen domains. In this paper, we propose a novel DG approach based on Deep Domain-Adversarial Image Generation (DDAIG). Specifically, DDAIG consists of three components, namely a label classifier, a domain classifier and a domain transformation network (DoTNet). The goal for DoTNet is to map the source training data to unseen domains. This is achieved by having a learning objective formulated to ensure that the generated data can be correctly classified by the label classifier while fooling the domain classifier. By augmenting the source training data with the generated unseen domain data, we can make the label classifier more robust to unknown domain changes. Extensive experiments on four DG datasets demonstrate the effectiveness of our approach.

Fine-grained sketch-based image retrieval (FG-SBIR) addresses the problem of retrieving a particular photo instance given a user's query sketch. Its widespread applicability is however hindered by the fact that drawing a sketch takes time, and most people struggle to draw a complete and faithful sketch. In this paper, we reformulate the conventional FG-SBIR framework to tackle these challenges, with the ultimate goal of retrieving the target photo with the least number of strokes possible. We further propose an on-the-fly design that starts retrieving as soon as the user starts drawing. To accomplish this, we devise a reinforcement learning based cross-modal retrieval framework that directly optimizes rank of the ground-truth photo over a complete sketch drawing episode. Additionally, we introduce a novel reward scheme that circumvents the problems related to irrelevant sketch strokes, and thus provides us with a more consistent rank list during the retrieval. We achieve superior early-retrieval efficiency over state-of-the-art methods and alternative baselines on two publicly available fine-grained sketch retrieval datasets.

A common strategy adopted by existing state-of-the-art unsupervised domain adaptation (UDA) methods is to employ two classifiers to identify the misaligned local regions between source and target domain. Following the 'wisdom of the crowd' principle, one has to ask: why stop at two? Indeed, we find that using more classifiers leads to better performance, but also introduces more model parameters, therefore risking overfitting. In this paper, we introduce a novel method called STochastic clAssifieRs (STAR) for addressing this problem. Instead of representing one classifier as a weight vector, STAR models it as a Gaussian distribution with its variance representing the inter-classifier discrepancy. With STAR, we can now sample an arbitrary number of classifiers from the distribution, whilst keeping the model size the same as having two classifiers. Extensive experiments demonstrate that a variety of existing UDA methods can greatly benefit from STAR and achieve the state-of-the-art performance on both image classification and semantic segmentation tasks.

ImageNet pre-training has long been considered crucial by the fine-grained sketch-based image retrieval (FG-SBIR) community due to the lack of large sketch-photo paired datasets for FG-SBIR training. In this paper, we propose a self-supervised alternative for representation pre-training. Specifically, we consider the jigsaw puzzle game of recomposing images from shuffled parts. We identify two key facets of jigsaw task design that are required for effective FG-SBIR pre-training. The first is formulating the puzzle in a mixed-modality fashion. Second we show that framing the optimisation as permutation matrix inference via Sinkhorn iterations is more effective than the common classifier formulation of Jigsaw self-supervision. Experiments show that this self-supervised pre-training strategy significantly outperforms the standard ImageNet-based pipeline across all four product-level FG-SBIR benchmarks. Interestingly it also leads to improved cross-category generalisation across both pre-train/fine-tune and fine-tune/testing stages.

Existing object detection methods typically rely on the availability of abundant labelled training samples per class and offline model training in a batch mode. These requirements substantially limit their scalability to open-ended accommodation of novel classes with limited labelled training data, both in terms of model accuracy and training efficiency during deployment. We present the first study aiming to go beyond these limitations by considering the Incremental Few-Shot Detection (iFSD) problem setting, where new classes must be registered incrementally (without revisiting base classes) and with few examples. To this end we propose OpeN-ended Centre nEt (ONCE), a detector designed for incrementally learning to detect novel class objects with few examples. This is achieved by an elegant adaptation of the efficient CentreNet detector to the few-shot learning scenario, and meta-learning a class-wise code generator model for registering novel classes. ONCE fully respects the incremental learning paradigm, with novel class registration requiring only a single forward pass of few-shot training samples, and no access to base classes - thus making it suitable for deployment on embedded devices, etc. Extensive experiments conducted on both the standard object detection (COCO, PASCAL VOC) and fashion landmark detection (DeepFashion2) tasks show the feasibility of iFSD for the first time, opening an interesting and very important line of research.

In recent years, visual question answering (VQA) has become topical. The premise of VQA's significance as a benchmark in AI, is that both the image and textual question need to be well understood and mutually grounded in order to infer the correct answer. However, current VQA models perhaps 'understand' less than initially hoped, and instead master the easier task of exploiting cues given away in the question and biases in the answer distribution [1]. In this paper we propose the inverse problem of VQA (iVQA). The iVQA task is to generate a question that corresponds to a given image and answer pair. We propose a variational iVQA model that can generate diverse, grammatically correct and content correlated questions that match the given answer. Based on this model, we show that iVQA is an interesting benchmark for visuo-linguistic understanding, and a more challenging alternative to VQA because an iVQA model needs to understand the image better to be successful. As a second contribution, we show how to use iVQA in a novel reinforcement learning framework to diagnose any existing VQA model by way of exposing its belief set: the set of question-answer pairs that the VQA model would predict true for a given image. This provides a completely new window into what VQA models 'believe' about images. We show that existing VQA models have more erroneous beliefs than previously thought, revealing their intrinsic weaknesses. Suggestions are then made on how to address these weaknesses going forward.

Person re-identification (re-id) aims to match people across non-overlapping camera views in a public space. This is a challenging problem because the people captured in surveillance videos often wear similar clothing. Consequently, the differences in their appearance are typically subtle and only detectable at particular locations and scales. In this paper, we propose a deep re-id network (MuDeep) that is composed of two novel types of layers - a multi-scale deep learning layer, and a leader-based attention learning layer. Specifically, the former learns deep discriminative feature representations at different scales, while the latter utilizes the information from multiple scales to lead and determine the optimal weightings for each scale. The importance of different spatial locations for extracting discriminative features is learned explicitly via our leader-based attention learning layer. Extensive experiments are carried out to demonstrate that the proposed MuDeep outperforms the state-of-the-art on a number of benchmarks and has a better generalization ability under a domain generalization setting.

During the past decade, both multi-label learning and zero-shot learning have attracted huge research attention, and significant progress has been made. Multi-label learning algorithms aim to predict multiple labels given one instance, while most existing zero-shot learning approaches target at predicting a single testing label for each unseen class via transferring knowledge from auxiliary seen classes to target unseen classes. However, relatively less effort has been made on predicting multiple labels in the zero-shot setting, which is nevertheless a quite challenging task. In this work, we investigate and formalize a flexible framework consisting of two components, i.e., visual-semantic embedding and zero-shot multi-label prediction. First, we present a deep regression model to project the visual features into the semantic space, which explicitly exploits the correlations in the intermediate semantic layer of word vectors and makes label prediction possible. Then, we formulate the label prediction problem as a pairwise one and employ Ranking SVM to seek the unique multi-label correlations in the embedding space. Furthermore, we provide a transductive multi-label zeroshot prediction approach that exploits the testing data manifold structure. We demonstrate the effectiveness of the proposed approach on three popular multi-label datasets with state-of-theart performance obtained on both conventional and generalized ZSL settings.

We present the first competitive drawing agent Pixelor that exhibits humanlevel performance at a Pictionary-like sketching game, where the participant whose sketch is recognized first is a winner. Our AI agent can autonomously sketch a given visual concept, and achieve a recognizable rendition as quickly or faster than a human competitor. The key to victory for the agent's goal is to learn the optimal stroke sequencing strategies that generate the most recognizable and distinguishable strokes first. Training Pixelor is done in two steps. First, we infer the stroke order that maximizes early recognizability of human training sketches. Second, this order is used to supervise the training of a sequence-to-sequence stroke generator. Our key technical contributions are a tractable search of the exponential space of orderings using neural sorting; and an improved Seq2Seq Wasserstein (S2S-WAE) generator that uses an optimal-transport loss to accommodate the multi-modal nature of the optimal stroke distribution. Our analysis shows that Pixelor is better than the human players of the Quick, Draw! game, under both AI and human judging of early recognition. To analyze the impact of human competitors' strategies, we conducted a further human study with participants being given unlimited thinking time and training in early recognizability by feedback from an AI judge. The study shows that humans do gradually improve their strategies with training, but overall Pixelor still matches human performance. The code and the dataset are available at http://sketchx.ai/pixelor.

In this paper, a unified approach is presented to transfer learning that addresses several source and target domain labelspace and annotation assumptions with a single model. It is particularly effective in handling a challenging case, where source and target label-spaces are disjoint, and outperforms alternatives in both unsupervised and semi-supervised settings. The key ingredient is a common representation termed Common Factorised Space. It is shared between source and target domains, and trained with an unsupervised factorisation loss and a graph-based loss. With a wide range of experiments, we demonstrate the flexibility, relevance and efficacy of our method, both in the challenging cases with disjoint label spaces, and in the more conventional cases such as unsupervised domain adaptation, where the source and target domains share the same label-sets.

Automated deception detection (ADD) from real-life videos is a challenging task. It specifically needs to address two problems: (1) Both face and body contain useful cues regarding whether a subject is deceptive. How to effectively fuse the two is thus key to the effectiveness of an ADD model. (2) Real-life deceptive samples are hard to collect; learning with limited training data thus challenges most deep learning based ADD models. In this work, both problems are addressed. Specifically, for face-body multimodal learning, a novel face-focused cross-stream network (FFCSN) is proposed. It differs significantly from the popular two-stream networks in that: (a) face detection is added into the spatial stream to capture the facial expressions explicitly, and (b) correlation learning is performed across the spatial and temporal streams for joint deep feature learning across both face and body. To address the training data scarcity problem, our FFCSN model is trained with both meta learning and adversarial learning. Extensive experiments show that our FFCSN model achieves state-of-the-art results. Further, the proposed FFCSN model as well as its robust training strategy are shown to be generally applicable to other human-centric video analysis tasks such as emotion recognition from user-generated videos.

Recently, large-scale few-shot learning (FSL) becomes topical. It is discovered that, for a large-scale FSL problem with 1,000 classes in the source domain, a strong baseline emerges, that is, simply training a deep feature embedding model using the aggregated source classes and performing nearest neighbor (NN) search using the learned features on the target classes. The state-of-the-art large-scale FSL methods struggle to beat this baseline, indicating intrinsic limitations on scalability. To overcome the challenge, we propose a novel large-scale FSL model by learning transferable visual features with the class hierarchy which encodes the semantic relations between source and target classes. Extensive experiments show that the proposed model significantly outperforms not only the NN baseline but also the state-of-the-art alternatives. Furthermore, we show that the proposed model can be easily extended to the large-scale zero-shot learning (ZSL) problem and also achieves the state-of-the-art results.

Fine-grained sketch-based image retrieval (FG-SBIR) addresses matching specific photo instance using free-hand sketch as a query modality. Existing models aim to learn an embedding space in which sketch and photo can be directly compared. While successful, they require instance-level pairing within each coarse-grained category as annotated training data. Since the learned embedding space is domain-specific, these models do not generalise well across categories. This limits the practical applicability of FG-SBIR. In this paper, we identify cross-category generalisation for FG-SBIR as a domain generalisation problem, and propose the first solution. Our key contribution is a novel unsupervised learning approach to model a universal manifold of prototypical visual sketch traits. This manifold can then be used to paramaterise the learning of a sketch/photo representation. Model adaptation to novel categories then becomes automatic via embedding the novel sketch in the manifold and updating the representation and retrieval function accordingly. Experiments on the two largest FG-SBIR datasets, Sketchy and QMUL-Shoe-V2, demonstrate the efficacy of our approach in enabling cross-category generalisation of FG-SBIR.

We aim to learn a domain generalizable person re-identification (ReID) model. When such a model is trained on a set of source domains (ReID datasets collected from different camera networks), it can be directly applied to any new unseen dataset for effective ReID without any model updating. Despite its practical value in real-world deployments, generalizable ReID has seldom been studied. In this work, a novel deep ReID model termed Domain-Invariant Mapping Network (DIMN) is proposed. DIMN is designed to learn a mapping between a person image and its identity classifier, i.e., it produces a classifier using a single shot. To make the model domain-invariant, we follow a meta-learning pipeline and sample a subset of source domain training tasks during each training episode. However, the model is significantly different from conventional meta-learning methods in that: (1) no model updating is required for the target domain, (2) different training tasks share a memory bank for maintaining both scalability and discrimination ability, and (3) it can be used to match an arbitrary number of identities in a target domain. Extensive experiments on a newly proposed large-scale ReID domain generalization benchmark show that our DIMN significantly outperforms alternative domain generalization or meta-learning methods.

In this paper, we propose to tackle the challenging few-shot learning (FSL) problem by learning global class representations using both base and novel class training samples. In each training episode, an episodic class mean computed from a support set is registered with the global representation via a registration module. This produces a registered global class representation for computing the classification loss using a query set. Though following a similar episodic training pipeline as existing meta learning based approaches, our method differs significantly in that novel class training samples are involved in the training from the beginning. To compensate for the lack of novel class training samples, an effective sample synthesis strategy is developed to avoid overfitting. Importantly, by joint base-novel class training, our approach can be easily extended to a more practical yet challenging FSL setting, i.e., generalized FSL, where the label space of test data is extended to both base and novel classes. Extensive experiments show that our approach is effective for both of the two FSL settings.

Automatic data abstraction is an important capability for both benchmarking machine intelligence and supporting summarization applications. In the former one asks whether a machine can `understand' enough about the meaning of input data to produce a meaningful but more compact abstraction. In the latter this capability is exploited for saving space or human time by summarizing the essence of input data. In this paper we study a general reinforcement learning based framework for learning to abstract sequential data in a goal-driven way. The ability to define different abstraction goals uniquely allows different aspects of the input data to be preserved according to the ultimate purpose of the abstraction. Our reinforcement learning objective does not require human-defined examples of ideal abstraction. Importantly our model processes the input sequence holistically without being constrained by the original input order. Our framework is also domain agnostic -- we demonstrate applications to sketch, video and text data and achieve promising results in all domains.

We aim to learn deep person re-identification (ReID) models that are robust against noisy training data. Two types of noise are prevalent in practice: (1) label noise caused by human annotator errors and (2) data outliers caused by person detector errors or occlusion. Both types of noise pose serious problems for training ReID models, yet have been largely ignored so far. In this paper, we propose a novel deep network termed DistributionNet for robust ReID. Instead of representing each person image as a feature vector, DistributionNet models it as a Gaussian distribution with its variance representing the uncertainty of the extracted features. A carefully designed loss is formulated in DistributionNet to unevenly allocate uncertainty across training samples. Consequently, noisy samples are assigned large variance/uncertainty, which effectively alleviates their negative impacts on model fitting. Extensive experiments demonstrate that our model is more effective than alternative noise-robust deep models. The source code is available at: https://github.com/TianyuanYu/DistributionNet.

As an instance-level recognition problem, person re-identification (ReID) relies on discriminative features, which not only capture different spatial scales but also encapsulate an arbitrary combination of multiple scales. We callse features of both homogeneous and heterogeneous scales omni-scale features. In this paper, a novel deep ReID CNN is designed, termed Omni-Scale Network (OSNet), for omni-scale feature learning. This is achieved by designing a residual block composed of multiple convolutional feature streams, each detecting features at a certain scale. Importantly, a novel unified aggregation gate is introduced to dynamically fuse multi-scale features with input-dependent channel-wise weights. To efficiently learn spatial-channel correlations and avoid overfitting, the building block uses both pointwise and depthwise convolutions. By stacking such blocks layer-by-layer, our OSNet is extremely lightweight and can be trained from scratch on existing ReID benchmarks. Despite its small model size, our OSNet achieves state-of-the-art performance on six person-ReID datasets. Code and models are available at: https://github.com/KaiyangZhou/deep-person-reid.

Human free-hand sketches provide the useful data for studying human perceptual grouping, where the grouping principles such as the Gestalt laws of grouping are naturally in play during both the perception and sketching stages. In this paper, we make the first attempt to develop a universal sketch perceptual grouper. That is, a grouper that can be applied to sketches of any category created with any drawing style and ability, to group constituent strokes/segments into semantically meaningful object parts. The first obstacle to achieving this goal is the lack of large-scale datasets with grouping annotation. To overcome this, we contribute the largest sketch perceptual grouping dataset to date, consisting of 20 000 unique sketches evenly distributed over 25 object categories. Furthermore, we propose a novel deep perceptual grouping model learned with both generative and discriminative losses. The generative loss improves the generalization ability of the model, while the discriminative loss guarantees both local and global grouping consistency. Extensive experiments demonstrate that the proposed grouper significantly outperforms the state-of-the-art competitors. In addition, we show that our grouper is useful for a number of sketch analysis tasks, including sketch semantic segmentation, synthesis, and fine-grained sketch-based image retrieval.

Image annotation aims to annotate a given image with a variable number of class labels corresponding to diverse visual concepts. In this paper, we address two main issues in large-scale image annotation: 1) how to learn a rich feature representation suitable for predicting a diverse set of visual concepts ranging from object, scene to abstract concept and 2) how to annotate an image with the optimal number of class labels. To address the first issue, we propose a novel multi-scale deep model for extracting rich and discriminative features capable of representing a wide range of visual concepts. Specifically, a novel two-branch deep neural network architecture is proposed, which comprises a very deep main network branch and a companion feature fusion network branch designed for fusing the multi-scale features computed from the main branch. The deep model is also made multi-modal by taking noisy user-provided tags as model input to complement the image input. For tackling the second issue, we introduce a label quantity prediction auxiliary task to the main label prediction task to explicitly estimate the optimal label number for a given image. Extensive experiments are carried out on two large-scale image annotation benchmark datasets, and the results show that our method significantly outperforms the state of the art.

Video summarization aims to facilitate large-scale video browsing by producing short, concise summaries that are diverse and representative of original videos. In this paper, we formulate video summarization as a sequential decision-making process and develop a deep summarization network (DSN) to summarize videos. DSN predicts for each video frame a probability, which indicates how likely a frame is selected, and then takes actions based on the probability distributions to select frames, forming video summaries. To train our DSN, we propose an end-to-end, reinforcement learning-based framework, where we design a novel reward function that jointly accounts for diversity and representativeness of generated summaries and does not rely on labels or user interactions at all. During training, the reward function judges how diverse and representative the generated summaries are, while DSN strives for earning higher rewards by learning to produce more diverse and more representative summaries. Since labels are not required, our method can be fully unsupervised. Extensive experiments on two benchmark datasets show that our unsupervised method not only outperforms other state-of-the-art unsupervised methods, but also is comparable to or even superior than most of published supervised approaches.

We propose a deep hashing framework for sketch retrieval that, for the first time, works on a multi-million scale human sketch dataset. Leveraging on this large dataset, we explore a few sketch-specific traits that were otherwise under-studied in prior literature. Instead of following the conventional sketch recognition task, we introduce the novel problem of sketch hashing retrieval which is not only more challenging, but also offers a better testbed for large-scale sketch analysis, since: (i) more fine-grained sketch feature learning is required to accommodate the large variations in style and Abstraction, and (ii) a compact binary code needs to be learned at the same time to enable efficient retrieval. Key to our network design is the embedding of unique characteristics of human sketch, where (i) a two-branch CNN-RNN architecture is adapted to explore the temporal ordering of strokes, and (ii) a novel hashing loss is specifically designed to accommodate both the temporal and Abstract traits of sketches. By working with a 3.8M sketch dataset, we show that state-of-the-art hashing models specifically engineered for static images fail to perform well on temporal sketch data. Our network on the other hand not only offers the best retrieval performance on various code sizes, but also yields the best generalization performance under a zero-shot setting and when re-purposed for sketch recognition. Such superior performances effectively demonstrate the benefit of our sketch-specific design.

Key to effective person re-identification (Re-ID) is modelling discriminative and view-invariant factors of person appearance at both high and low semantic levels. Recently developed deep Re-ID models either learn a holistic single semantic level feature representation and/or require laborious human annotation of these factors as attributes. We propose Multi-Level Factorisation Net (MLFN), a novel network architecture that factorises the visual appearance of a person into latent discriminative factors at multiple semantic levels without manual annotation. MLFN is composed of multiple stacked blocks. Each block contains multiple factor modules to model latent factors at a specific level, and factor selection modules that dynamically select the factor modules to interpret the content of each input image. The outputs of the factor selection modules also provide a compact latent factor descriptor that is complementary to the conventional deeply learned features. MLFN achieves state-of-the-art results on three Re-ID datasets, as well as compelling results on the general object categorisation CIFAR-100 dataset.

Recently the widely used multi-view learning model, Canonical Correlation Analysis (CCA) has been generalised to the non-linear setting via deep neural networks. Existing deep CCA models typically first decorrelate the feature dimensions of each view before the different views are maximally correlated in a common latent space. This feature decorrelation is achieved by enforcing an exact decorrelation constraint; these models are thus computationally expensive due to the matrix inversion or SVD operations required for exact decorrelation at each training iteration. Furthermore, the decorrelation step is often separated from the gradient descent based optimisation, resulting in sub-optimal solutions. We propose a novel deep CCA model Soft CCA to overcome these problems. Specifically, exact decorrelation is replaced by soft decorrelation via a mini-batch based Stochastic Decorrelation Loss (SDL) to be optimised jointly with the other training objectives. Extensive experiments show that the proposed soft CCA is more effective and efficient than existing deep CCA models. In addition, our SDL loss can be applied to other deep models beyond multi-view learning, and obtains superior performance compared to existing decorrelation losses.

Contemporary deep learning techniques have made image recognition a reasonably reliable technology. However training effective photo classifiers typically takes numerous examples which limits image recognition's scalability and applicability to scenarios where images may not be available. This has motivated investigation into zero-shot learning, which addresses the issue via knowledge transfer from other modalities such as text. In this paper we investigate an alternative approach of synthesizing image classifiers: almost directly from a user's imagination, via freehand sketch. This approach doesn't require the category to be nameable or describable via attributes as per zero-shot learning. We achieve this via training a model regression network to map from free-hand sketch space to the space of photo classifiers. It turns out that this mapping can be learned in a category-agnostic way, allowing photo classifiers for new categories to be synthesized by user with no need for annotated training photos. We also demonstrate that this modality of classifier generation can also be used to enhance the granularity of an existing photo classifier, or as a complement to name-based zero-shot learning.

We propose the inverse problem of Visual question answering (iVQA), and explore its suitability as a benchmark for visuo-linguistic understanding. The iVQA task is to generate a question that corresponds to a given image and answer pair. Since the answers are less informative than the questions, and the questions have less learnable bias, an iVQA model needs to better understand the image to be successful than a VQA model. We pose question generation as a multi-modal dynamic inference process and propose an iVQA model that can gradually adjust its focus of attention guided by both a partially generated question and the answer. For evaluation, apart from existing linguistic metrics, we propose a new ranking metric. This metric compares the ground truth question's rank among a list of distractors, which allows the drawbacks of different algorithms and sources of error to be studied. Experimental results show that our model can generate diverse, grammatically correct and content correlated questions that match the given answer.

To see is to sketch - free-hand sketching naturally builds ties between human and machine vision. In this paper, we present a novel approach for translating an object photo to a sketch, mimicking the human sketching process. This is an extremely challenging task because the photo and sketch domains differ significantly. Furthermore, human sketches exhibit various levels of sophistication and abstraction even when depicting the same object instance in a reference photo. This means that even if photo-sketch pairs are available, they only provide weak supervision signal to learn a translation model. Compared with existing supervised approaches that solve the problem of D(E(photo)) → sketch), where E(·) and D(·) denote encoder and decoder respectively, we take advantage of the inverse problem (e.g., D(E(sketch) → photo), and combine with the unsupervised learning tasks of within-domain reconstruction, all within a multi-task learning framework. Compared with existing unsupervised approaches based on cycle consistency (i.e., D(E(D(E(photo)))) → photo), we introduce a shortcut consistency enforced at the encoder bottleneck (e.g., D(E(photo)) → photo) to exploit the additional self-supervision. Both qualitative and quantitative results show that the proposed model is superior to a number of state-of-the-art alternatives. We also show that the synthetic sketches can be used to train a better fine-grained sketch-based image retrieval (FG-SBIR) model, effectively alleviating the problem of sketch data scarcity.

We present a conceptually simple, flexible, and general framework for few-shot learning, where a classifier must learn to recognise new classes given only few examples from each. Our method, called the Relation Network (RN), is trained end-to-end from scratch. During meta-learning, it learns to learn a deep distance metric to compare a small number of images within episodes, each of which is designed to simulate the few-shot setting. Once trained, a RN is able to classify images of new classes by computing relation scores between query images and the few examples of each new class without further updating the network. Besides providing improved performance on few-shot learning, our framework is easily extended to zero-shot learning. Extensive experiments on five benchmarks demonstrate that our simple approach provides a unified and effective approach for both of these two tasks.

Model distillation is an effective and widely used technique to transfer knowledge from a teacher to a student network. The typical application is to transfer from a powerful large network or ensemble to a small network, in order to meet the low-memory or fast execution requirements. In this paper, we present a deep mutual learning (DML) strategy. Different from the one-way transfer between a static pre-defined teacher and a student in model distillation, with DML, an ensemble of students learn collaboratively and teach each other throughout the training process. Our experiments show that a variety of network architectures benefit from mutual learning and achieve compelling results on both category and instance recognition tasks. Surprisingly, it is revealed that no prior powerful teacher network is necessary - mutual learning of a collection of simple student networks works, and moreover outperforms distillation from a more powerful yet static teacher.

Zero-shot learning (ZSL) aims to recognize unseen object classes without any training samples, which can be regarded as a form of transfer learning from seen classes to unseen ones. This is made possible by learning a projection between a feature space and a semantic space (e.g. attribute space). Key to ZSL is thus to learn a projection function that is robust against the often large domain gap between the seen and unseen classes. In this paper, we propose a novel ZSL model termed domain-invariant projection learning (DIPL). Our model has two novel components: (1) A domain-invariant feature self-reconstruction task is introduced to the seen/unseen class data, resulting in a simple linear formulation that casts ZSL into a min-min optimization problem. Solving the problem is non-trivial, and a novel iterative algorithm is formulated as the solver, with rigorous theoretic algorithm analysis provided. (2) To further align the two domains via the learned projection, shared semantic structure among seen and unseen classes is explored via forming superclasses in the semantic space. Extensive experiments show that our model outperforms the state-of-the-art alternatives by significant margins.

Most existing person re-identification (re-id) methods are unsuitable for real-world deployment due to two reasons: Unscalability to large population size , and Inadaptability over time . In this work, we present a unified solution to address both problems. Specifically, we propose to construct an identity regression space (IRS) based on embedding different training person identities (classes) and formulate re-id as a regression problem solved by identity regression in the IRS. The IRS approach is characterised by a closed-form solution with high learning efficiency and an inherent incremental learning capability with human-in-the-loop. Extensive experiments on four benchmarking datasets (VIPeR, CUHK01, CUHK03 and Market-1501) show that the IRS model not only outperforms state-of-the-art re-id methods, but also is more scalable to large re-id population size by rapidly updating model and actively selecting informative samples with reduced human labelling effort.

Zero-Shot Learning (ZSL) for visual recognition is typically achieved by exploiting a semantic embedding space. In such a space, both seen and unseen class labels as well as image features can be embedded so that the similarity among them can be measured directly. In this work, we consider that the key to effective ZSL is to compute an optimal distance metric in the semantic embedding space. Existing ZSL works employ either euclidean or cosine distances. However, in a high-dimensional space where the projected class labels (prototypes) are sparse, these distances are suboptimal, resulting in a number of problems including hubness and domain shift. To overcome these problems, a novel manifold distance computed on a semantic class prototype graph is proposed which takes into account the rich intrinsic semantic structure, i.e., semantic manifold, of the class prototype distribution. To further alleviate the domain shift problem, a new regularisation term is introduced into a ranking loss based embedding model. Specifically, the ranking loss objective is regularised by unseen class prototypes to prevent the projected object features from being biased towards the seen prototypes. Extensive experiments on four benchmarks show that our method significantly outperforms the state-of-the-art.

A number of vision problems such as zero-shot learning and person re-identification can be considered as cross-class transfer learning problems. As mid-level semantic properties shared cross different object classes, attributes have been studied extensively for knowledge transfer across classes. Most previous attribute learning methods focus only on human-defined/nameable semantic attributes, whilst ignoring the fact there also exist undefined/latent shareable visual properties, or latent attributes. These latent attributes can be either discriminative or non-discriminative parts depending on whether they can contribute to an object recognition task. In this work, we argue that learning the latent attributes jointly with user-defined semantic attributes not only leads to better representation but also helps semantic attribute prediction. A novel dictionary learning model is proposed which decomposes the dictionary space into three parts corresponding to semantic, latent discriminative and latent background attributes respectively. Such a joint attribute learning model is then extended by following a multi-task transfer learning framework to address a more challenging unsupervised domain adaptation problem, where annotations are only available on an auxiliary dataset and the target dataset is completely unlabelled. Extensive experiments show that the proposed models, though being linear and thus extremely efficient to compute, produce state-of-the-art results on both zero-shot learning and person re-identification.

Existing zero-shot learning (ZSL) models typically learn a projection function from a feature space to a semantic embedding space (e.g. attribute space). However, such a projection function is only concerned with predicting the training seen class semantic representation (e.g. attribute prediction) or classification. When applied to test data, which in the context of ZSL contains different (unseen) classes without training data, a ZSL model typically suffers from the project domain shift problem. In this work, we present a novel solution to ZSL based on learning a Semantic AutoEncoder (SAE). Taking the encoder-decoder paradigm, an encoder aims to project a visual feature vector into the semantic space as in the existing ZSL models. However, the decoder exerts an additional constraint, that is, the projection/code must be able to reconstruct the original visual feature. We show that with this additional reconstruction constraint, the learned projection function from the seen classes is able to generalise better to the new unseen classes. Importantly, the encoder and decoder are linear and symmetric which enable us to develop an extremely efficient learning algorithm. Extensive experiments on six benchmark datasets demonstrate that the proposed SAE outperforms significantly the existing ZSL models with the additional benefit of lower computational cost. Furthermore, when the SAE is applied to supervised clustering problem, it also beats the state-of-the-art.

Zero-shot learning (ZSL) models rely on learning a joint embedding space where both textual/semantic description of object classes and visual representation of object images can be projected to for nearest neighbour search. Despite the success of deep neural networks that learn an end-to-end model between text and images in other vision problems such as image captioning, very few deep ZSL model exists and they show little advantage over ZSL models that utilise deep feature representations but do not learn an end-to-end embedding. In this paper we argue that the key to make deep ZSL models succeed is to choose the right embedding space. Instead of embedding into a semantic space or an intermediate space, we propose to use the visual space as the embedding space. This is because that in this space, the subsequent nearest neighbour search would suffer much less from the hubness problem and thus become more effective. This model design also provides a natural mechanism for multiple semantic modalities (e.g.,~attributes and sentence descriptions) to be fused and optimised jointly in an end-to-end manner. Extensive experiments on four benchmarks show that our model significantly outperforms the existing models.

Person Re-identification (re-id) aims to match people across non-overlapping camera views in a public space. It is a challenging problem because many people captured in surveillance videos wear similar clothes. Consequently, the differences in their appearance are often subtle and only detectable at the right location and scales. Existing re-id models, particularly the recently proposed deep learning based ones match people at a single scale. In contrast, in this paper, a novel multi-scale deep learning model is proposed. Our model is able to learn deep discriminative feature representations at different scales and automatically determine the most suitable scales for matching. The importance of different spatial locations for extracting discriminative features is also learned explicitly. Experiments are carried out to demonstrate that the proposed model outperforms the state-of-the art on a number of benchmarks.

Human sketches are unique in being able to capture both the spatial topology of a visual object, as well as its subtle appearance details. Fine-grained sketch-based image retrieval (FG-SBIR) importantly leverages on such fine-grained characteristics of sketches to conduct instance-level retrieval of photos. Nevertheless, human sketches are often highly abstract and iconic, resulting in severe misalignments with candidate photos which in turn make subtle visual detail matching difficult. Existing FG-SBIR approaches focus only on coarse holistic matching via deep cross-domain representation learning, yet ignore explicitly accounting for fine-grained details and their spatial context. In this paper, a novel deep FG-SBIR model is proposed which differs significantly from the existing models in that: (1) It is spatially aware, achieved by introducing an attention module that is sensitive to the spatial position of visual details: (2) It combines coarse and fine semantic information via a shortcut connection fusion block: and (3) It models feature correlation and is robust to misalignments between the extracted features across the two domains by introducing a novel higher-order learnable energy function (HOLEF) based loss. Extensive experiments show that the proposed deep spatial-semantic attention model significantly outperforms the state-of-the-art.

Modeling hairstyles for classification, synthesis and image editing has many practical applications. However, existing hairstyle datasets, such as the Beauty e-Expert dataset, are too small for developing and evaluating computer vision models, especially the recent deep generative models such as generative adversarial network (GAN). In this paper, we contribute a new large-scale hairstyle dataset called Hairstyle30k, which is composed of 30k images containing 64 different types of hairstyles. To enable automated generating and modifying hairstyles in images, we also propose a novel GAN model termed Hairstyle GAN (H-GAN) which can be learned efficiently. Extensive experiments on the new dataset as well as existing benchmark datasets demonstrate the effectiveness of proposed H-GAN model

A weakly supervised semantic segmentation (WSSS) method aims to learn a segmentation model from weak (image-level) as opposed to strong (pixel-level) labels. By avoiding the tedious pixel-level annotation process, it can exploit the unlimited supply of user-tagged images from media-sharing sites such as Flickr for large scale applications. However, these `free' tags/labels are often noisy and few existing works address the problem of learning with both weak and noisy labels. In this work, we cast the WSSS problem into a label noise reduction problem. Specifically, after segmenting each image into a set of superpixels, the weak and potentially noisy image-level labels are propagated to the superpixel level resulting in highly noisy labels; the key to semantic segmentation is thus to identify and correct the superpixel noisy labels. To this end, a novel L1-optimisation based sparse learning model is formulated to directly and explicitly detect noisy labels. To solve the L1-optimisation problem, we further develop an efficient learning algorithm by introducing an intermediate labelling variable. Extensive experiments on three benchmark datasets show that our method yields state-of-the-art results given noise-free labels, whilst significantly outperforming the existing methods when the weak labels are also noisy.

We propose to model complex visual scenes using a non-parametric Bayesian model learned from weakly labelled images abundant on media sharing sites such as Flickr. Given weak image-level annotations of objects and attributes without locations or associations between them, our model aims to learn the appearance of object and attribute classes as well as their association on each object instance. Once learned, given an image, our model can be deployed to tackle a number of vision problems in a joint and coherent manner, including recognising objects in the scene (automatic object annotation), describing objects using their attributes (attribute prediction and association), and localising and delineating the objects (object detection and semantic segmentation). This is achieved by developing a novel Weakly Supervised Markov Random Field Stacked Indian Buffet Process (WS-MRF-SIBP) that models objects and attributes as latent factors and explicitly captures their correlations within and across superpixels. Extensive experiments on benchmark datasets demonstrate that our weakly supervised model significantly outperforms weakly supervised alternatives and is often comparable with existing strongly supervised models on a variety of tasks including semantic segmentation, automatic image annotation and retrieval based on object-attribute associations.

Due to the rapid technological development of various sensors, a huge volume of high spatial resolution (HSR) image data can now be acquired. How to efficiently recognize the scenes from such HSR image data has become a critical task. Conventional approaches to remote sensing scene classification only utilize information from HSR images. Therefore, they always need a large amount of labeled data and cannot recognize the images from an unseen scene class without any visual sample in the labeled data. To overcome this drawback, we propose a novel approach for recognizing images from unseen scene classes, i.e., zero-shot scene classification (ZSSC). In this approach, we first use the well-known natural language process model, word2vec, to map names of seen/unseen scene classes to semantic vectors. A semantic-directed graph is then constructed over the semantic vectors for describing the relationships between unseen classes and seen classes. To transfer knowledge from the images in seen classes to those in unseen classes, we make an initial label prediction on test images by an unsupervised domain adaptation model. With the semantic-directed graph and initial prediction, a label-propagation algorithm is then developed for ZSSC. By leveraging the visual similarity among images from the same scene class, a label refinement approach based on sparse learning is used to suppress the noise in the zero-shot classification results. Experimental results show that the proposed approach significantly outperforms the state-of-the-art approaches in ZSSC.

We study the problem of fine-grained sketch-based image retrieval. By performing instance-level (rather than category-level) retrieval, it embodies a timely and practical application, particularly with the ubiquitous availability of touchscreens. Three factors contribute to the challenging nature of the problem: 1) free-hand sketches are inherently abstract and iconic, making visual comparisons with photos difficult; 2) sketches and photos are in two different visual domains, i.e., black and white lines versus color pixels; and 3) fine-grained distinctions are especially challenging when executed across domain and abstraction-level. To address these challenges, we propose to bridge the image-sketch gap both at the high level via parts and attributes, as well as at the low level via introducing a new domain alignment method. More specifically, first, we contribute a data set with 304 photos and 912 sketches, where each sketch and image is annotated with its semantic parts and associated part-level attributes. With the help of this data set, second, we investigate how strongly supervised deformable part-based models can be learned that subsequently enable automatic detection of part-level attributes, and provide pose-aligned sketch-image comparisons. To reduce the sketch-image gap when comparing low-level features, third, we also propose a novel method for instance-level domain-alignment that exploits both subspace and instance-level cues to better align the domains. Finally, fourth, these are combined in a matching framework integrating aligned low-level features, mid-level geometric structure, and high-level semantic attributes. Extensive experiments conducted on our new data set demonstrate effectiveness of the proposed method.

Modeling complex crowd behaviour for tasks such as rare event detection has received increasing interest. However, existing methods are limited because (1) they are sensitive to noise often resulting in a large number of false alarms; and (2) they rely on elaborate models leading to high computational cost thus unsuitable for processing a large number of video inputs in real-time. In this paper, we overcome these limitations by introducing a novel complex behaviour modeling framework, which consists of a Binarized Cumulative Directional (BCD) feature as representation, novel spatial and temporal context modeling via an iterative correlation maximization, and a set of behaviour models, each being a simple Bernoulli distribution. Despite its simplicity, our experiments on three benchmark datasets show that it significantly outperforms the state-of-the-art for both temporal video segmentation and rare event detection. Importantly, it is extremely efficient — reaches 90Hz on a normal PC platform using MATLAB.

Most existing person re-identification (Re-ID) approaches follow a supervised learning framework, in which a large number of labelled matching pairs are required for training. This severely limits their scalability in realworld applications. To overcome this limitation, we develop a novel cross-dataset transfer learning approach to learn a discriminative representation. It is unsupervised in the sense that the target dataset is completely unlabelled. Specifically, we present an multi-task dictionary learning method which is able to learn a dataset-shared but target-data-biased representation. Experimental results on five benchmark datasets demonstrate that the method significantly outperforms the state-of-the-art.

We investigate the problem of fine-grained sketch-based image retrieval (SBIR), where free-hand human sketches are used as queries to perform instance-level retrieval of images. This is an extremely challenging task because (i) visual comparisons not only need to be fine-grained but also executed cross-domain, (ii) free-hand (finger) sketches are highly abstract, making fine-grained matching harder, and most importantly (iii) annotated cross-domain sketch-photo datasets required for training are scarce, challenging many state-of-the-art machine learning techniques. In this paper, for the first time, we address all these challenges, providing a step towards the capabilities that would underpin a commercial sketch-based image retrieval application. We introduce a new database of 1,432 sketchphoto pairs from two categories with 32,000 fine-grained triplet ranking annotations. We then develop a deep tripletranking model for instance-level SBIR with a novel data augmentation and staged pre-training strategy to alleviate the issue of insufficient fine-grained training data. Extensive experiments are carried out to contribute a variety of insights into the challenges of data sufficiency and over-fitting avoidance when training deep networks for finegrained cross-domain ranking tasks.

Most existing person re-identification (re-id) methods focus on learning the optimal distance metrics across camera views. Typically a person's appearance is represented using features of thousands of dimensions, whilst only hundreds of training samples are available due to the difficulties in collecting matched training images. With the number of training samples much smaller than the feature dimension, the existing methods thus face the classic small sample size (SSS) problem and have to resort to dimensionality reduction techniques and/or matrix regularisation, which lead to loss of discriminative power. In this work, we propose to overcome the SSS problem in re-id distance metric learning by matching people in a discriminative null space of the training data. In this null space, images of the same person are collapsed into a single point thus minimising the within-class scatter to the extreme and maximising the relative between-class separation simultaneously. Importantly, it has a fixed dimension, a closed-form solution and is very efficient to compute. Extensive experiments carried out on five person re-identification benchmarks including VIPeR, PRID2011, CUHK01, CUHK03 and Market1501 show that such a simple approach beats the state-of-the-art alternatives, often by a big margin.

Zero shot learning (ZSL) provides a solution to recognising unseen classes without class labelled data for model learning. Most ZSL methods aim to learn a mapping from a visual feature space to a semantic embedding space, e.g. attribute or word vector spaces. The use of word vector space is particularly attractive as compared to attribute, it offers vast auxiliary classes with free parts embedding without human annotation. However, using the word vector embedding often provides weaker discriminative power than manually labelled attributes of the auxiliary classes. This is compounded further in zero-shot action recognition due to richer content variations among action classes. In this work we propose to explore a broader semantic contextual information in the text domain to enrich the word vector representation of action classes. We show through extensive experiments that this method improves significantly the performance of a number of existing word vector embedding ZSL methods. Moreover, it also outperforms attribute embedding ZSL with human annotation.

Most existing person re-identification (ReID) methods assume the availability of extensively labelled cross-view person pairs and a closed-set scenario (i.e. all the probe people exist in the gallery set). These two assumptions significantly limit their usefulness and scalability in real-world applications, particularly with large scale camera networks. To overcome the limitations, we introduce a more challenging yet realistic ReID setting termed OneShot-OpenSet-RelD, and propose a novel Regularised Kernel Subspace Learning model for ReID under this setting. Our model differs significantly from existing ReID methods due to its ability of effectively learning cross-view identity-specific information from unlabelled data alone, and its flexibility of naturally accommodating pairwise labels if available.

The problem of estimating subjective visual properties from image and video has attracted increasing interest. A subjective visual property is useful either on its own (e.g. image and video interestingness) or as an intermediate representation for visual recognition (e.g. a relative attribute). Due to its ambiguous nature, annotating the value of a subjective visual property for learning a prediction model is challenging. To make the annotation more reliable, recent studies employ crowdsourcing tools to collect pairwise comparison labels. However, using crowdsourced data also introduces outliers. Existing methods rely on majority voting to prune the annotation outliers/errors. They thus require a large amount of pairwise labels to be collected. More importantly as a local outlier detection method, majority voting is ineffective in identifying outliers that can cause global ranking inconsistencies. In this paper, we propose a more principled way to identify annotation outliers by formulating the subjective visual property prediction task as a unified robust learning to rank problem, tackling both the outlier detection and learning to rank jointly. This differs from existing methods in that (1) the proposed method integrates local pairwise comparison labels together to minimise a cost that corresponds to global inconsistency of ranking order, and (2) the outlier detection and learning to rank problems are solved jointly. This not only leads to better detection of annotation outliers but also enables learning with extremely sparse annotations.

Solving the problem of matching people across non-overlapping multi-camera views, known as person re-identification (re-id), has received increasing interests in computer vision. In a real-world application scenario, a watch-list (gallery set) of a handful of known target people are provided with very few (in many cases only a single) image(s) (shots) per target. Existing re-id methods are largely unsuitable to address this open-world re-id challenge because they are designed for (1) a closed-world scenario where the gallery and probe sets are assumed to contain exactly the same people, (2) person-wise identification whereby the model attempts to verify exhaustively against each individual in the gallery set, and (3) learning a matching model using multi-shots. In this paper, a novel transfer local relative distance comparison (t-LRDC) model is formulated to address the open-world person re-identification problem by one-shot group-based verification. The model is designed to mine and transfer useful information from a labelled open-world non-target dataset. Extensive experiments demonstrate that the proposed approach outperforms both non-transfer learning and existing transfer learning based re-id methods.

Human action can be recognized from a single still image by modeling human-object interactions (HOIs), which infers the mutual spatial structure information between human and the manipulated object as well as their appearance. Existing approaches rely heavily on accurate detection of human and object and estimation of human pose; they are thus sensitive to large variations of human poses, occlusion, and unsatisfactory detection of small size objects. To overcome this limitation, a novel exemplar-based approach is proposed in this paper. Our approach learns a set of spatial pose-object interaction exemplars, which are probabilistic density functions describing spatially how a person is interacting with a manipulated object for different activities. Specifically, a new framework consisting of an exemplar-based HOI descriptor and an associated matching model is formulated for robust human action recognition in still images. In addition, the framework is extended to perform HOI recognition in videos, where the proposed exemplar representation is used for implicit frame selection to negate irrelevant or noisy frames by temporal structured HOI modeling. Extensive experiments are carried out on two image action datasets and two video action datasets. The results demonstrate the effectiveness of our proposed methods and show that our approach is able to achieve state-of-the-art performance, compared with several recently proposed competitors.

Many existing approaches to person re-identification (Re-ID) are based on supervised learning, which requires hundreds of matching pairs to be labelled for each pair of cameras. This severely limits their scalability for real-world applications. This work aims to overcome this limitation by developing a novel unsupervised Re-ID approach. The approach is based on a new dictionary learning for sparse coding formulation with a graph Laplacian regularisation term whose value is set iteratively. As an unsupervised model, the dictionary learning model is well-suited to the unsupervised task, whilst the regularisation term enables the exploitation of cross-view identity-discriminative information ignored by existing unsupervised Re-ID methods. Importantly this model is also flexible in utilising any labelled data if available. Experiments on two benchmark datasets demonstrate that the proposed approach significantly outperforms the state-of-the-arts.

We propose a multi-scale multi-channel deep neural network framework that, for the first time, yields sketch recognition performance surpassing that of humans. Our superior performance is a result of explicitly embedding the unique characteristics of sketches in our model: (i) a network architecture designed for sketch rather than natural photo statistics, (ii) a multi-channel generalisation that encodes sequential ordering in the sketching process, and (iii) a multi-scale network ensemble with joint Bayesian fusion that accounts for the different levels of abstraction exhibited in free-hand sketches. We show that state-of-the-art deep networks specifically engineered for photos of natural objects fail to perform well on sketch recognition, regardless whether they are trained using photo or sketch. Our network on the other hand not only delivers the best performance on the largest human sketch dataset to date, but also is small in size making efficient training possible using just CPUs.

We propose a perceptual grouping framework that organizes image edges into meaningful structures and demonstrate its usefulness on various computer vision tasks. Our grouper formulates edge grouping as a graph partition problem, where a learning to rank method is developed to encode probabilities of candidate edge pairs. In particular, RankSVM is employed for the first time to combine multiple Gestalt principles as cue for edge grouping. Afterwards, an edge grouping based object proposal measure is introduced that yields proposals comparable to state-of-the-art alternatives. We further show how human-like sketches can be generated from edge groupings and consequently used to deliver state-of-the-art sketch-based image retrieval performance. Last but not least, we tackle the problem of freehand human sketch segmentation by utilizing the proposed grouper to cluster strokes into semantic object parts.

Learning semantic attributes for person re-identification and description-based person search has gained increasing interest due to attributes' great potential as a pose and view-invariant representation. However, existing attribute-centric approaches have thus far underperformed state-of-the-art conventional approaches. This is due to their nonscalable need for extensive domain (camera) specific annotation. In this paper we present a new semantic attribute learning approach for person re-identification and search. Our model is trained on existing fashion photography datasets - either weakly or strongly labelled. It can then be transferred and adapted to provide a powerful semantic description of surveillance person detections, without requiring any surveillance domain supervision. The resulting representation is useful for both unsupervised and supervised person re-identification, achieving state-of-the-art and near state-of-the-art performance respectively. Furthermore, as a semantic representation it allows description-based person search to be integrated within the same framework.

Zero-shot learning (ZSL) can be considered as a special case of transfer learning where the source and target domains have different tasks/label spaces and the target domain is unlabelled, providing little guidance for the knowledge transfer. A ZSL method typically assumes that the two domains share a common semantic representation space, where a visual feature vector extracted from an image/video can be projected/embedded using a projection function. Existing approaches learn the projection function from the source domain and apply it without adaptation to the target domain. They are thus based on naive knowledge transfer and the learned projections are prone to the domain shift problem. In this paper a novel ZSL method is proposed based on unsupervised domain adaptation. Specifically, we formulate a novel regularised sparse coding framework which uses the target domain class labels' projections in the semantic space to regularise the learned target domain projection thus effectively overcoming the projection domain shift problem. Extensive experiments on four object and action recognition benchmark datasets show that the proposed ZSL method significantly outperforms the state-of-the-arts.

In real world person re-identification (re-id), images of people captured at very different resolutions from different locations need be matched. Existing re-id models typically normalise all person images to the same size. However, a low-resolution (LR) image contains much less information about a person, and direct image scaling and simple size normalisation as done in conventional re-id methods cannot compensate for the loss of information. To solve this LR person re-id problem, we propose a novel joint multi-scale learning framework, termed joint multi-scale discriminant component analysis (JUDEA). The key component of this framework is a heterogeneous class mean discrepancy (HCMD) criterion for cross-scale image domain alignment, which is optimised simultaneously with discriminant modelling across multiple scales in the joint learning framework. Our experiments show that the proposed JUDEA framework outperforms existing representative re-id methods as well as other related LR visual matching models applied for the LR person re-id problem.

We address a new partial person re-identification (re-id) problem, where only a partial observation of a person is available for matching across different non-overlapping camera views. This differs significantly from the conventional person re-id setting where it is assumed that the full body of a person is detected and aligned. To solve this more challenging and realistic re-id problem without the implicit assumption of manual body-parts alignment, we propose a matching framework consisting of 1) a local patch-level matching model based on a novel sparse representation classification formulation with explicit patch ambiguity modelling, and 2) a global part-based matching model providing complementary spatial layout information. Our framework is evaluated on a new partial person re-id dataset as well as two existing datasets modified to include partial person images. The results show that the proposed method outperforms significantly existing re-id methods as well as other partial visual matching methods.

Most existing zero-shot learning approaches exploit transfer learning via an intermediate semantic representation shared between an annotated auxiliary dataset and a target dataset with different classes and no annotation. A projection from a low-level feature space to the semantic representation space is learned from the auxiliary dataset and applied without adaptation to the target dataset. In this paper we identify two inherent limitations with these approaches. First, due to having disjoint and potentially unrelated classes, the projection functions learned from the auxiliary dataset/domain are biased when applied directly to the target dataset/domain. We call this problem the projection domain shift problem and propose a novel framework, transductive multi-view embedding, to solve it. The second limitation is the prototype sparsity problem which refers to the fact that for each target class, only a single prototype is available for zero-shot learning given a semantic representation. To overcome this problem, a novel heterogeneous multi-view hypergraph label propagation method is formulated for zero-shot learning in the transductive embedding space. It effectively exploits the complementary information offered by different semantic representations and takes advantage of the manifold structures of multiple representation spaces in a coherent manner. We demonstrate through extensive experiments that the proposed approach (1) rectifies the projection shift between the auxiliary and target domains, (2) exploits the complementarity of multiple semantic representations, (3) significantly outperforms existing methods for both zero-shot and N-shot recognition on three image and video benchmark datasets, and (4) enables novel cross-view annotation tasks.

We address the problem of localisation of objects as bounding boxes in images and videos with weak labels. This weakly supervised object localisation problem has been tackled in the past using discriminative models where each object class is localised independently from other classes. In this paper, a novel framework based on Bayesian joint topic modelling is proposed, which differs significantly from the existing ones in that: (1) All foreground object classes are modelled jointly in a single generative model that encodes multiple object co-existence so that "explaining away" inference can resolve ambiguity and lead to better learning and localisation. (2) Image backgrounds are shared across classes to better learn varying surroundings and "push out" objects of interest. (3) Our model can be learned with a mixture of weakly labelled and unlabelled data, allowing the large volume of unlabelled images on the Internet to be exploited for learning. Moreover, the Bayesian formulation enables the exploitation of various types of prior knowledge to compensate for the limited supervision offered by weakly labelled data, as well as Bayesian domain adaptation for transfer learning. Extensive experiments on the PASCAL VOC, ImageNet and YouTube-Object videos datasets demonstrate the effectiveness of our Bayesian joint model for weakly supervised object localisation.

The rapid development of social media sharing has created a huge demand for automatic media classification and annotation techniques. Attribute learning has emerged as a promising paradigm for bridging the semantic gap and addressing data sparsity via transferring attribute knowledge in object recognition and relatively simple action classification. In this paper, we address the task of attribute learning for understanding multimedia data with sparse and incomplete labels. In particular, we focus on videos of social group activities, which are particularly challenging and topical examples of this task because of their multimodal content and complex and unstructured nature relative to the density of annotations. To solve this problem, we 1) introduce a concept of semilatent attribute space, expressing user-defined and latent attributes in a unified framework, and 2) propose a novel scalable probabilistic topic model for learning multimodal semilatent attributes, which dramatically reduces requirements for an exhaustive accurate attribute ontology and expensive annotation effort. We show that our framework is able to exploit latent attributes to outperform contemporary approaches for addressing a variety of realistic multimedia sparse data learning tasks including: multitask learning, learning with label noise, N-shot transfer learning, and importantly zero-shot learning.

Video analysis often begins with background subtraction. This problem is often approached in two steps-a background model followed by a regularisation scheme. A model of the background allows it to be distinguished on a per-pixel basis from the foreground, whilst the regularisation combines information from adjacent pixels. We present a new method based on Dirichlet process Gaussian mixture models, which are used to estimate per-pixel background distributions. It is followed by probabilistic regularisation. Using a non-parametric Bayesian method allows per-pixel mode counts to be automatically inferred, avoiding over-/under- fitting. We also develop novel model learning algorithms for continuous update of the model in a principled fashion as the scene changes. These key advantages enable us to outperform the state-of-the-art alternatives on four benchmarks.

A number of computer vision problems such as human age estimation, crowd density estimation and body/face pose (view angle) estimation can be formulated as a regression problem by learning a mapping function between a high dimensional vector-formed feature input and a scalar-valued output. Such a learning problem is made difficult due to sparse and imbalanced training data and large feature variations caused by both uncertain viewing conditions and intrinsic ambiguities between observable visual features and the scalar values to be estimated. Encouraged by the recent success in using attributes for solving classification problems with sparse training data, this paper introduces a novel cumulative attribute concept for learning a regression model when only sparse and imbalanced data are available. More precisely, low-level visual features extracted from sparse and imbalanced image samples are mapped onto a cumulative attribute space where each dimension has clearly defined semantic interpretation (a label) that captures how the scalar output value (e.g. age, people count) changes continuously and cumulatively. Extensive experiments show that our cumulative attribute framework gains notable advantage on accuracy for both age estimation and crowd counting when compared against conventional regression models, especially when the labelled training data is sparse with imbalanced sampling.

We propose a principled probabilistic formulation of object saliency as a sampling problem. This novel formulation allows us to learn, from a large corpus of unlabelled images, which patches of an image are of the greatest interest and most likely to correspond to an object. We then sample the object saliency map to propose object locations. We show that using only a single object location proposal per image, we are able to correctly select an object in over 42% of the images in the Pascal VOC 2007 dataset, substantially outperforming existing approaches. Furthermore, we show that our object proposal can be used as a simple unsupervised approach to the weakly supervised annotation problem. Our simple unsupervised approach to annotating objects of interest in images achieves a higher annotation accuracy than most weakly supervised approaches.

Human action can be recognised from a single still image by modelling Human-object interaction (HOI), which infers the mutual spatial structure information between human and object as well as their appearance. Existing approaches rely heavily on accurate detection of human and object, and estimation of human pose. They are thus sensitive to large variations of human poses, occlusion and unsatisfactory detection of small size objects. To overcome this limitation, a novel exemplar based approach is proposed in this work. Our approach learns a set of spatial pose-object interaction exemplars, which are density functions describing how a person is interacting with a manipulated object for different activities spatially in a probabilistic way. A representation based on our HOI exemplar thus has great potential for being robust to the errors in human/object detection and pose estimation. A new framework consists of a proposed exemplar based HOI descriptor and an activity specific matching model that learns the parameters is formulated for robust human activity recognition. Experiments on two benchmark activity datasets demonstrate that the proposed approach obtains state-of-the-art performance.

Regression-based techniques have shown promising results for people counting in crowded scenes. However, most existing techniques require expensive and laborious data annotation for model training. In this study, we propose to address this problem from three perspectives: (1) Instead of exhaustively annotating every single frame, the most informative frames are selected for annotation automatically and actively. (2) Rather than learning from only labelled data, the abundant unlabelled data are exploited. (3) Labelled data from other scenes are employed to further alleviate the burden for data annotation. All three ideas are implemented in a unified active and semi-supervised regression framework with ability to perform transfer learning, by exploiting the underlying geometric structure of crowd patterns via manifold analysis. Extensive experiments validate the effectiveness of our approach.

We address the problem of localisation of objects as bounding boxes in images with weak labels. This weakly supervised object localisation problem has been tackled in the past using discriminative models where each object class is localised independently from other classes. We propose a novel framework based on Bayesian joint topic modelling. Our framework has three distinctive advantages over previous works: (1) All object classes and image backgrounds are modelled jointly together in a single generative model so that "explaining away" inference can resolve ambiguity and lead to better learning and localisation. (2) The Bayesian formulation of the model enables easy integration of prior knowledge about object appearance to compensate for limited supervision. (3) Our model can be learned with a mixture of weakly labelled and unlabelled data, allowing the large volume of unlabelled images on the Internet to be exploited for learning. Extensive experiments on the challenging VOC dataset demonstrate that our approach outperforms the state-of-the-art competitors.

Sketch is used for rendering the visual world since prehistoric times, and has become ubiquitous nowadays with the increasing availability of touchscreens on portable devices. However, how to automatically map images to sketches, a problem that has profound implications on applications such as sketch-based image retrieval, still remains open. In this paper, we propose a novel method that draws a sketch automatically from a single natural image. Sketch extraction is posed within an unified contour grouping framework, where perceptual grouping is first used to form contour segment groups, followed by a group-based contour simplification method that generate the final sketches. In our experiment, for the first time we pose sketch evaluation as a sketch-based object recognition problem and the results validate the effectiveness of our system over the state-of-the-arts alternatives.

Gestalt principles, a set of conjoining rules derived from human visual studies, have been known to play an important role in computer vision. Many applications such as image segmentation, contour grouping and scene understanding often rely on such rules to work. However, the problem of Gestalt confliction, i.e., the relative importance of each rule compared with another, remains unsolved. In this paper, we investigate the problem of perceptual grouping by quantifying the confliction among three commonly used rules: similarity, continuity and proximity. More specifically, we propose to quantify the importance of Gestalt rules by solving a learning to rank problem, and formulate a multi-label graph-cuts algorithm to group image primitives while taking into account the learned Gestalt confliction. Our experiment results confirm the existence of Gestalt confliction in perceptual grouping and demonstrate an improved performance when such a confliction is accounted for via the proposed grouping algorithm. Finally, a novel cross domain image classification method is proposed by exploiting perceptual grouping as representation.

Matching people across nonoverlapping camera views at different locations and different times, known as person reidentification, is both a hard and important problem for associating behavior of people observed in a large distributed space over a prolonged period of time. Person reidentification is fundamentally challenging because of the large visual appearance changes caused by variations in view angle, lighting, background clutter, and occlusion. To address these challenges, most previous approaches aim to model and extract distinctive and reliable visual features. However, seeking an optimal and robust similarity measure that quantifies a wide range of features against realistic viewing conditions from a distance is still an open and unsolved problem for person reidentification. In this paper, we formulate person reidentification as a relative distance comparison (RDC) learning problem in order to learn the optimal similarity measure between a pair of person images. This approach avoids treating all features indiscriminately and does not assume the existence of some universally distinctive and reliable features. To that end, a novel relative distance comparison model is introduced. The model is formulated to maximize the likelihood of a pair of true matches having a relatively smaller distance than that of a wrong match pair in a soft discriminant manner. Moreover, in order to maintain the tractability of the model in large scale learning, we further develop an ensemble RDC model. Extensive experiments on three publicly available benchmarking datasets are carried out to demonstrate the clear superiority of the proposed RDC models over related popular person reidentification techniques. The results also show that the new RDC models are more robust against visual appearance changes and less susceptible to model overfitting compared to other related existing models.

Discovering rare categories and classifying new instances of them are important data mining issues in many fields, but fully supervised learning of a rare class classifier is prohibitively costly in labeling effort. There has therefore been increasing interest both in active discovery: to identify new classes quickly, and active learning: to train classifiers with minimal supervision. These goals occur together in practice and are intrinsically related because examples of each class are required to train a classifier. Nevertheless, very few studies have tried to optimise them together, meaning that data mining for rare classes in new domains makes inefficient use of human supervision. Developing active learning algorithms to optimise both rare class discovery and classification simultaneously is challenging because discovery and classification have conflicting requirements in query criteria. In this paper, we address these issues with two contributions: a unified active learning model to jointly discover new categories and learn to classify them by adapting query criteria online; and a classifier combination algorithm that switches generative and discriminative classifiers as learning progresses. Extensive evaluation on a batch of standard UCI and vision data sets demonstrates the superiority of this approach over existing methods.

Most existing approaches to training object detectors rely on fully supervised learning, which requires the tedious manual annotation of object location in a training set. Recently there has been an increasing interest in developing weakly supervised approach to detector training where the object location is not manually annotated but automatically determined based on binary (weak) labels indicating if a training image contains the object. This is a challenging problem because each image can contain many candidate object locations which partially overlaps the object of interest. Existing approaches focus on how to best utilise the binary labels for object location annotation. In this paper we propose to solve this problem from a very different perspective by casting it as a transfer learning problem. Specifically, we formulate a novel transfer learning based on learning to rank, which effectively transfers a model for automatic annotation of object location from an auxiliary dataset to a target dataset with completely unrelated object categories. We show that our approach outperforms existing state-of-the-art weakly supervised approach to annotating objects in the challenging VOC dataset.

Learning from streams of evolving and unbounded data is an important problem, for example in visual surveillance or internet scale data. For such large and evolving real-world data, exhaustive supervision is impractical, particularly so when the full space of classes is not known in advance therefore joint class discovery (exploration) and boundary learning (exploitation) becomes critical. Active learning has shown promise in jointly optimising exploration-exploitation with minimal human supervision. However, existing active learning methods either rely on heuristic multi-criteria weighting or are limited to batch processing. In this paper, we present a new unified framework for joint exploration-exploitation active learning in streams without any heuristic weighting. Extensive evaluation on classification of various image and surveillance video datasets demonstrates the superiority of our framework over existing methods.

Solving the person re-identification problem has become important for understanding people's behaviours in a multicamera network of non-overlapping views. In this work, we address the problem of re-identification from a set-based verification perspective. More specifically, we have a small set of target people on a watch list (a set) and we aim to verify whether a query image of a person is on this watch list. This differs from the existing person re-identification problem in that the probe is verified against a small set of known people but requires much higher degree of verification accuracy with very limited sampling data for each candidate in the set. That is, rather than recognising everybody in the scene, we consider identifying a small set of target people against non-target people when there is only a limited number of target training samples and a large number of unlabelled (unknown) non-target samples available. To this end, we formulate a transfer learning framework for mining discriminant information from non-target people data to solve the watch list set verification problem. Based on the proposed approach, we introduce the concepts of multi-shot and one-shot verifications. We also design new criteria for evaluating the performance of the proposed transfer learning method against the i-LIDS and ETHZ data sets.

Activity modeling and unusual event detection in a network of cameras is challenging, particularly when the camera views are not overlapped. We show that it is possible to detect unusual events in multiple disjoint cameras as context-incoherent patterns through incremental learning of time delayed dependencies between distributed local activities observed within and across camera views. Specifically, we model multicamera activities using a Time Delayed Probabilistic Graphical Model (TD-PGM) with different nodes representing activities in different decomposed regions from different views and the directed links between nodes encoding their time delayed dependencies. To deal with visual context changes, we formulate a novel incremental learning method for modeling time delayed dependencies that change over time. We validate the effectiveness of the proposed approach using a synthetic data set and videos captured from a camera network installed at a busy underground station.

Context is critical for reducing the uncertainty in object detection. However, context modeling is challenging because there are often many different types of contextual information coexisting with different degrees of relevance to the detection of target object(s) in different images. It is therefore crucial to devise a context model to automatically quantify and select the most effective contextual information for assisting in detecting the target object. Nevertheless, the diversity of contextual information means that learning a robust context model requires a larger training set than learning the target object appearance model, which may not be available in practice. In this work, a novel context modeling framework is proposed without the need for any prior scene segmentation or context annotation. We formulate a polar geometric context descriptor for representing multiple types of contextual information. In order to quantify context, we propose a new maximum margin context (MMC) model to evaluate and measure the usefulness of contextual information directly and explicitly through a discriminant context inference method. Furthermore, to address the problem of context learning with limited data, we exploit the idea of transfer learning based on the observation that although two categories of objects can have very different visual appearance, there can be similarity in their context and/or the way contextual information helps to distinguish target objects from nontarget objects. To that end, two novel context transfer learning models are proposed which utilize training samples from source object classes to improve the learning of the context model for a target object class based on a joint maximum margin learning framework. Experiments are carried out on PASCAL VOC2005 and VOC2007 data sets, a luggage detection data set extracted from the i-LIDS data set, and a vehicle detection data set extracted from outdoor surveillance footage. Our results validate the effectiveness of the proposed models for quantifying and transferring contextual information, and demonstrate that they outperform related alternative context models.

Real-world classification problems, such as visual surveillance and network intrusion detection, often contain common yet uninteresting background classes and rare but interesting classes, that need to be both discovered and classified. Active learning offers a suitable solution to joint rare class discovery and classification, by minimising the manual labelling of training data. A novel active learning approach is proposed, which automatically balances the competing goals of new class discovery and improving classification. Crucially it is free of tuneable parameters. Using Dirichlet processes a new active learning criterion is formulated, based on first computing the probability that unlabelled exemplars are from a new class, in addition to existing classes, and subsequently the probability of misclassification, which is then used for query selection. The proposed approach works with any probabilistic classification model and its effectiveness is demonstrated on multiple problems.

The detection of human action in videos of busy natural scenes with dynamic background is of interest for applications such as video surveillance. Taking a conventional fully supervised approach, the spatio-temporal locations of the action of interest have to be manually annotated frame by frame in the training videos, which is tedious and unreliable. In this paper, for the first time, a weakly supervised action detection method is proposed which only requires binary labels of the videos indicating the presence of the action of interest. Given a training set of binary labelled videos, the weakly supervised learning (WSL) problem is recast as a multiple instance learning (MIL) problem. A novel MIL algorithm is developed which differs from the existing MIL algorithms in that it locates the action of interest spatially and temporally by globally optimising both interand intra-class distance. We demonstrate through experiments that our WSL approach can achieve comparable detection performance to a fully supervised learning approach, and that the proposed MIL algorithm significantly outperforms the existing ones.

Matching people across non-overlapping camera views, known as person re-identification, is challenging due to the lack of spatial and temporal constraints and large visual appearance changes caused by variations in view angle, lighting, background clutter and occlusion. To address these challenges, most previous approaches aim to extract visual features that are both distinctive and stable under appearance changes. However, most visual features and their combinations under realistic conditions are neither stable nor distinctive thus should not be used indiscriminately. In this paper, we propose to formulate person re-identification as a distance learning problem, which aims to learn the optimal distance that can maximises matching accuracy regardless the choice of representation. To that end, we introduce a novel Probabilistic Relative Distance Comparison (PRDC) model, which differs from most existing distance learning methods in that, rather than minimising intra-class variation whilst maximising intra-class variation, it aims to maximise the probability of a pair of true match having a smaller distance than that of a wrong match pair. This makes our model more tolerant to appearance changes and less susceptible to model over-fitting. Extensive experiments are carried out to demonstrate that 1) by formulating the person re-identification problem as a distance learning problem, notable improvement on matching accuracy can be obtained against conventional person re-identification techniques, which is particularly significant when the training sample size is small; and 2) our PRDC outperforms not only existing distance learning methods but also alternative learning methods based on boosting and learning to rank.

In the security domain a key problem is identifying rare behaviours of interest. Training examples for these behaviours may or may not exist, and if they do exist there will be few examples, quite probably one. We present a novel weakly supervised algorithm that can detect behaviours that either have never before been seen or for which there are few examples. Global context is modelled, allowing the detection of abnormal behaviours that in isolation appear normal. Pragmatic aspects are considered, such that no parameter tuning is required and real time performance is achieved.

A conventional approach to learning object detectors uses fully supervised learning techniques which assumes that a training image set with manual annotation of object bounding boxes are provided. The manual annotation of objects in large image sets is tedious and unreliable. Therefore, a weakly supervised learning approach is desirable, where the training set needs only binary labels regarding whether an image contains the target object class. In the weakly supervised approach a detector is used to iteratively annotate the training set and learn the object model. We present a novel weakly supervised learning framework for learning an object detector. Our framework incorporates a new initial annotation model to start the iterative learning of a detector and a model drift detection method that is able to detect and stop the iterative learning when the detector starts to drift away from the objects of interest. We demonstrate the effectiveness of our approach on the challenging PASCAL 2007 dataset.

Providing methods to support semantic interaction with growing volumes of video data is an increasingly important challenge for data mining. To this end, there has been some success in recognition of simple objects and actions in video, however most of this work requires strongly supervised training data. The supervision cost of these approaches therefore renders them economically non-scalable for real world applications. In this paper we address the problem of learning to annotate and retrieve semantic tags of human actions in realistic video data with sparsely provided tags of semantically salient activities. This is challenging because of (1) the multi-label nature of the learning problem and (2) realistic videos are often dominated by (semantically uninteresting) background activity un-supported by any tags of interest, leading to a strong irrelevant data problem. To address these challenges, we introduce a new topic model based approach to video tag annotation. Our model simultaneously learns a low dimensional representation of the video data, which dimensions are semantically relevant (supported by tags), and how to annotate videos with tags. Experimental evaluation on three different video action/activity datasets demonstrate the challenge of this problem, and value of our contribution.

One of the most interesting and desired capabilities for automated video behavior analysis is the identification of rarely occurring and subtle behaviors. This is of practical value because dangerous or illegal activities often have few or possibly only one prior example to learn from and are often subtle. Rare and subtle behavior learning is challenging for two reasons: (1) Contemporary modeling approaches require more data and supervision than may be available and (2) the most interesting and potentially critical rare behaviors are often visually subtle-occurring among more obvious typical behaviors or being defined by only small spatio-temporal deviations from typical behaviors. In this paper, we introduce a novel weakly supervised joint topic model which addresses these issues. Specifically, we introduce a multiclass topic model with partially shared latent structure and associated learning and inference algorithms. These contributions will permit modeling of behaviors from as few as one example, even without localization by the user and when occurring in clutter, and subsequent classification and localization of such behaviors online and in real time. We extensively validate our approach on two standard public-space data sets, where it clearly outperforms a batch of contemporary alternatives.

Among various factors that can affect the performance of gait recognition, changes in viewpoint pose the biggest problem. In this work, we develop a novel approach to cross-view gait recognition with the view angle of a probe gait sequence unknown. We formulate a Gaussian Process (GP) classification framework to estimate the view angle of each probe gait sequence. To measure the similarity of gait sequences captured at different view angles, we model the correlation of gait sequences from different views using Canonical Correlation Analysis (CCA) and use the correlation strength as similarity measure. This differs significantly from existing approaches, which reconstruct gait features in different views either through 2D view transformation or 3D calibration. Without explicit reconstruction, our approach can cope with feature mis-match across view and is more robust against feature noise. Our experiments validate that the proposed method significantly outperforms the existing state-of-the-art methods.

Problem This paper addresses the problem of recognising realistic human actions captured in unconstrained environments (Fig. 1). Existing approaches for action recognition have been focused on improving visual feature representation using either spatio-temporal interest points or key-points trajectories. However, these methods are insufficient to handle the situations when action videos are recorded in unconstrained environments because: (1) Reliable visual features are hard to be extracted due to occlusions, illumination change, scale variation and background clutters. (2) Effectiveness of visual features are strongly dependent on the unpredictable characteristics of camera movements. (3) Complicated visual actions result in unequal discriminativeness of visual features. Our Solutions In this paper, we present a novel framework for recognising realistic human actions in unconstrained environments. The novelties of our work lie in three aspects: First, we propose a new action representation based on computing a rich set of descriptors from key point trajectories. Second, in order to cope with drastic changes in motion characteristics with and without camera movements, we develop an adaptive feature fusion method to combine different local motion descriptors for improving model robustness against feature noise and background clutters. Finally, we propose a novel Multi-Class Delta Latent Dirichlet Allocation (MC-∆LDA) model for feature selection. The most informative features in a high dimensional feature space are selected collaboratively rather than independently. Motion Descriptors We first compute trajectories of key-points using KLT tracker and SIFT matching. After trajectory pruning by identifying the Region of Interest (ROI), we compute three types of motion descriptors from the survived trajectories. First, Orientation-Magnitude Descriptor is extracted by quantising orientation and magnitude of motion between two consecutive points in the same trajectory. Second, Trajectory Shape Descriptor is extracted by computing Fourier coefficients of a single trajectory. Finally, Appearance Descriptor is extracted by computing the SIFT features at all points of a trajectory. Interest Point Features We also detect spatio-temporal interest points as they contain complementary information to trajectory features. At an interest point, a surrounding 3D cuboid is extracted. We use gradient vectors to describe these cuboids and PCA to reduce descriptor’s dimensionality. Adaptive Feature Fusion We wish to fuse adaptively trajectory based descriptors with 3D interest point based descriptors according the presence of camera movement. The presence of moving camera is detected by computing the global optical flow over all frames in a clip. If the majority of the frames contain global motion, we regard the clip as being recorded by a moving camera. For clips without camera movement, both interest point and trajectory based descriptors can be computed reliably and thus both types of descriptors are used for recognition. In contrast, when camera motion can be detected, interest point based descriptors are less meaningful so only trajectory descriptors are employed. Collaborative Feature Selection We propose a MC-∆LDA model (Fig. 2) for collaboratively selecting dominant features for classification. We consider each video clip x j is a mixture of Nt topics Φ = {φt}t t=1 (to be discovered), each of which φt is a multinomial distribution over Nw words (visual features). The MC-∆LDA model aims to constrain topic proportion non-uniformly and on a per-clip basis. For each video clip belonging to action category Ac, we model it as a mixture of: (1) Ns t topics which are shared by all Nc category of actions, and (2) Nt,c topics which are uniquely associated with action category Ac. In MC-∆LDA, the nonuniform proportion of topic mixture for a single clip x j is enforced by its action class label c j and the hyperparameter αc for the corresponding action class c. Given the total number of topics Nt = Ns t +∑ Nc c=1 Nt,c, the structure of the MC-∆LDA model, and the observable variables (clips x j and action labels c j), we can learn the Ns t shared topics as well as all ∑c c=1 Nt,c unique topics for all Nc classes of actions. We use the N s t topics shared by all actions for selecting discriminative features. The Ns t shared topics are represented as an Nw×N t dimension matrix Φs. The feature selection can be summarised into two steps: (1) For each feature vk, k = Figure 1: Actions captured in an unconstrained environments, YouTube dataset. From left to right: cycling, diving, soccer juggling, and walking with a dog.

Solving the person re-identification problem involves matching observations of individuals across disjoint camera views. The problem becomes particularly hard in a busy public scene as the number of possible matches is very high. This is further compounded by significant appearance changes due to varying lighting conditions, viewing angles and body poses across camera views. To address this problem, existing approaches focus on extracting or learning discriminative features followed by template matching using a distance measure. The novelty of this work is that we reformulate the person reidentification problem as a ranking problem and learn a subspace where the potential true match is given highest ranking rather than any direct distance measure. By doing so, we convert the person re-identification problem from an absolute scoring problem to a relative ranking problem. We further develop an novel Ensemble RankSVM to overcome the scalability limitation problem suffered by existing SVM-based ranking methods. This new model reduces significantly memory usage therefore is much more scalable, whilst maintaining high-level performance. We present extensive experiments to demonstrate the performance gain of the proposed ranking approach over existing template matching and classification models. 1

For automated surveillance, it is useful to detect specific actions performed by people in busy natural environments. This differs from and thus more challenging than the intensively studied action recognition problem in that for action detection in crowd an action of interest is often overwhelmed by large number of background activities of other objects in the scene. Motivated by the success of sliding-window based 2D object detection approaches, in this paper, we propose to tackle the problem by learning a discriminative classifier from annotated 3D action cuboids to capture the intra-class variation, and sliding 3D search windows for detection. The key innovation of our method is a novel greedy k nearest neighbour algorithm for automated annotation of positive training data, by which an action detector can be learned with only a single training sequence being annotated thus greatly alleviating the tedious and unreliable 3D manual annotation. Extensive experiments on real-world action detection datasets demonstrate that our detector trained with minimal annotation can achieve comparable results to that learned with full annotation, and outperforms existing methods.

Gait is characterised by the relative motions between different body parts during walking. However, most recently proposed gait representation approaches such as Gait Energy Image (GEI) and Motion Silhouettes Image (MSI) capture only the motion intensity information whilst ignoring the equally important but less reliable information about the direction of relative motion. They thus essentially sacrifice discriminative power in exchange for robustness. In this paper, we propose a novel gait representation based on optical flow fields computed from normalized and centred person images over a complete gait cycle. In our representation, both the motion intensity and the motion direction information is captured in a set of motion descriptors. To achieve robustness against noise, instead of relying on the exact value of the flow vectors, the flow direction is discretised and a histogram based direction representation is formulated. Compared to the existing model-free gait representations, our representation is not only more discriminative, but also less sensitive to changes in various covariate conditions including clothing, carrying, shoe, and speed. Extensive experiments on both indoor and outdoor public datasets have been carried out to demonstrate that our representation outperforms the state-of-the-art.

We present a new approach for activity modelling and anomaly detection based on non-parametric Gaussian Process (GP) models. Specifically, GP regression models are formulated to learn non-linear relationships between multi-object activity patterns observed from semantically decomposed regions in complex scenes. Predictive distributions are inferred from the regression models to compare with the actual observations for real-time anomaly detection. The use of a flexible, non-parametric model alleviates the difficult problem of selecting appropriate model complexity encountered in parametric models such as Dynamic Bayesian Networks (DBNs). Crucially, our GP models need fewer parameters; they are thus less likely to overfit given sparse data. In addition, our approach is robust to the inevitable noise in activity representation as noise is modelled explicitly in the GP models. Experimental results on a public traffic scene show that our models outperform DBNs in terms of anomaly sensitivity, noise robustness, and flexibility in modelling complex activity.

We propose a novel technique for head pose classification in crowded public space under poor lighting and in low-resolution video images. Unlike previous approaches, we avoid the need for explicit segmentation of skin and hair regions from a head image and implicitly encode spatial information using a grid map for more robustness given lowresolution images. Specifically, a new head pose descriptor is formulated using similarity distance maps by indexing each pixel of a head image to the mean appearance templates of head images at different poses. These distance feature maps are then used to train a multi-class Support Vector Machine for pose classification. Our approach is evaluated against established techniques [3, 13, 14] using the i-LIDS underground scene dataset [9] under challenging lighting and viewing conditions. The results demonstrate that our model gives significant improvement in head pose estimation accuracy, with over 80% pose recognition rate against 32% from the best of existing models.

Tracking is regarded as one of the most fundamental tasks in computer vision. It is used in many computer vision applications in fields such as surveillance, robotic navigation and 3D reconstruction to name but a few. Despite decades of research, the goal of fully automatic tracking of arbitrary types of objects in real world conditions is still an open problem. In this paper, we take a step toward the goal of general real-world tracking, and demonstrate a unified generative model for Bayesian multifeature, adaptive target tracking, or AMFT for short (Adaptive Multiple Feature Tracker). We derive a unified generative model for multi-sensory adaptive tracking which cleanly integrates tracking and the modeling of appearance change across multiple features in the same framework. The unified multi-feature observation model ensures that if one feature is not confident, e.g., color after an object crosses into a region of shadow, it is automatically down-weighted in its contribution to the appearance model update. In this way, without pre-training of specific object models, we achieve an extensible tracker for general object types, robust to real-world problems of clutter, appearance/lighting change and target model drift. The standard modeling assumptions made by a non-adaptive generative model are illustrated by the probabilistic graphical model in Figure 1(a). The unknown target state (e.g., location, size, velocity) xt is assumed to change with time t according to some process parameterized by A. At every time t, we make some noisy observations zt of the target xt (e.g., raw image or color histograms). The target is then tracked online by computing the posterior, p(xt |z1:t) over the true target location recursively. In the case of the Kalman filter (KF), all the distributions involved are Gaussian. In the case of the particle filter (PF), all the distributions involved are represented non-parametrically by a set of samples [1]. The true target model, e.g., the appearance or color histogram to search for, is assumed to be part of the parameters H, i.e., it is known and fixed by an operator or initialized by some external process. In many cases however, the true appearance of the target H may change significantly in time, e.g., the appearance changes when a subject moves between shade and sunlight. This is the case for outdoor surveillance applications and is the motivation for this research. Adaptive trackers [2, 3, 4, 6] have been proposed to update the target appearance online in various heuristic ways. We can formalise this more general modeling assumption generatively, by the generalized dynamic Bayesian network illustrated in Figure 1(b). In contrast to Figure 1(a), the true target model which was previously included in the fixed parameters H, is now included as the the initial condition y0 of a dynamic latent variable yt , formalizing the modeling assumption that the target appearance can change over time. In addition to the target state xt , the target appearance yt will therefore be incrementally and recursively updated as part of the process of inferring the latent variables in this model p(xt ,yt |z1:t). The latent space is of course now greatly expanded, and poses a more challenging inference problem than that of Figure 1(a). In Section 2 of the paper, we detail the specific parametric form of the model and an efficient inference algorithm. We evaluate our method (AMFT) against three contemporary trackers: A standard single feature particle filter (PF), mean-shift (MS) [5] and incremental visual tracking (IVT) [4]. The PF and MS trackers are non-adaptive color-based trackers, while IVT aims for pose and illumination change robustness by performing online adaptation in a subspace appearance model. Note that the AMFT, PF and IVT trackers track object scale, but MS does not. We evaluated these methods on a series of challenging video clips exhibiting a wide variety of data and object types for tracking, including far-field indoor and outdoor pedestrians with and without carried objects, vehicle tracking, and near-field indoor face trackH 1 x2 x3

In a crowded public space, people often walk in groups, either with people they know or strangers. Associating a group of people over space and time can assist understanding individual’s behaviours as it provides vital visual context for matching individuals within the group. Seemingly an ‘easier’ task compared with person matching given more and richer visual content, this problem is in fact very challenging because a group of people can be highly non-rigid with changing relative position of people within the group and severe self-occlusions. In this paper, for the first time, the problem of matching/associating groups of people over large space and time captured in multiple non-overlapping camera views is addressed. Specifically, a novel people group representation and a group matching algorithm are proposed. The former addresses changes in the relative positions of people in a group and the latter deals with variations in illumination and viewpoint across camera views. In addition, we demonstrate a notable enhancement on individual person matching by utilising the group description as visual context. Our methods are validated using the 2008 i-LIDS Multiple-Camera Tracking Scenario (MCTS) dataset on multiple camera views from a busy airport arrival hall.

Much of recent action recognition research is based on space-time interest points extracted from video using a Bag of Words (BOW) representation. It mainly relies on the discriminative power of individual local space-time descriptors, whilst ignoring potentially valuable information about the global spatio-temporal distribution of interest points. In this paper, we propose a novel action recognition approach which differs significantly from previous interest points based approaches in that only the global spatiotemporal distribution of the interest points are exploited. This is achieved through extracting holistic features from clouds of interest points accumulated over multiple temporal scales followed by automatic feature selection. Our approach avoids the non-trivial problems of selecting the optimal space-time descriptor, clustering algorithm for constructing a codebook, and selecting codebook size faced by previous interest points based methods. Our model is able to capture smooth motions, robust to view changes and occlusions at a low computation cost. Experiments using the KTH and WEIZMANN datasets demonstrate that our approach outperforms most existing methods.

We propose a novel approach for modelling correlations between activities in a busy public space captured by multiple non-overlapping and uncalibrated cameras. In our approach, each camera view is automatically decomposed into semantic regions, across which different spatio-temporal activity patterns are observed. A novel Cross Canonical Correlation Analysis (xCCA) framework is formulated to detect and quantify temporal and causal relationships between regional activities within and across camera views. The approach accomplishes three tasks: (1) estimate the spatial and temporal topology of the camera network; (2) facilitate more robust and accurate person re-identification; (3) perform global activity modelling and video temporal segmentation by linking visual evidence collected across camera views. Our approach differs from the state of the art in that it does not rely on either intra or inter camera tracking. It therefore can be applied to even the most challenging video surveillance settings featured with severe occlusions and extremely low spatial and temporal resolutions. Its effectiveness is demonstrated using 153 hours of videos from 8 cameras installed in a busy underground station.

This paper addresses the problem of fully automated mining of public space video data. A novel Markov Clustering Topic Model (MCTM) is introduced which builds on existing Dynamic Bayesian Network models (e.g. HMMs) and Bayesian topic models (e.g. Latent Dirichlet Allocation), and overcomes their drawbacks on accuracy, robustness and computational efficiency. Specifically, our model profiles complex dynamic scenes by robustly clustering visual events into activities and these activities into global behaviours, and correlates behaviours over time. A collapsed Gibbs sampler is derived for offline learning with unlabeled training data, and significantly, a new approximation to online Bayesian inference is formulated to enable dynamic scene understanding and behaviour mining in new video data online in real-time. The strength of this model is demonstrated by unsupervised learning of dynamic scene models, mining behaviours and detecting salient events in three complex and crowded public scenes.

We present a novel approach for detecting global behaviour anomalies in multiple disjoint cameras by learning time delayed dependencies between activities cross camera views. Specifically, we propose to model multi-camera activities using a Time Delayed Probabilistic Graphical Model (TD-PGM) with different nodes representing activities in different semantically decomposed regions from different camera views, and the directed links between nodes encoding causal relationships between the activities. A novel two-stage structure learning algorithm is formulated to learn globally optimised time-delayed dependencies. A new cumulative abnormality score is also introduced to replace the conventional log-likelihood score for gaining significantly more robust and reliable real-time anomaly detection. The effectiveness of the proposed approach is validated using a camera network installed at a busy underground station.

Context is critical for minimising ambiguity in object detection. In this work, a novel context modelling framework is proposed without the need of any prior scene segmentation or context annotation. This is achieved by exploring a new polar geometric histogram descriptor for context representation. In order to quantify context, we formulate a new context risk function and a maximum margin context (MMC) model to solve the minimization problem of the risk function. Crucially, the usefulness and goodness of contextual information is evaluated directly and explicitly through a discriminant context inference method and a context confidence function, so that only reliable contextual information that is relevant to object detection is utilised. Experiments on PASCAL VOC2005 and i-LIDS datasets demonstrate that the proposed context modelling approach improves object detection significantly and outperforms a state-of-the-art alternative context model.

The strength of gait, compared to other biometrics, is that it does not require cooperative subjects. Previoius gait recognition approaches were evaluated using a gallery set consisting of gait sequences of people under similar covariate conditions (i.e. clothing, surface, carrying, and view conditions). This evaluation procedure, however, implies that the gait data are collected in a cooperative manner so that the covariate conditions are known a priori. In this work, the performance of state of the art gait recognition approaches are evaluated without the assumption on cooperative subjects, i.e. the gallery set consists of a mixture of gait sequences under different unknown covariate conditions. The results show that the performance of the existing approaches drop drastically under this more realistic experimental setup. We argue that selecting the most relevant gait features that are invariant to changes in gait covariate conditions is the key to develop a gait recognition system that works without subject cooperation. To that end, we propose a novel gait recognition approach, which performs automatic feature selection on each pair gallery and probe gait sequences, and seamlessly integrates feature selection with an Adaptive Component and Discriminant Analysis (ACDA) for fast recognition. Experiments are carried out to demonstrate that the proposed approach significantly outperforms the existing techniques. 1

We present a novel framework for inferring global behaviour patterns through modelling behaviour correlations in a wide-area scene and detecting any anomaly in behaviours occurring both locally and globally. Specifically, we propose a semantic scene segmentation model to decompose a wide-area scene into regions where behaviours share similar characteristic and are represented as classes of video events bearing similar features. To model behavioural correlations globally, we investigate both a probabilistic Latent Semantic Analysis (pLSA) model and a two-stage hierarchical pLSA model for global behaviour inference and anomaly detection. The proposed framework is validated by experiments using complex crowded outdoor scenes. 1

The appearance of individuals captured by multiple non-overlapping cameras varies greatly due to pose and illumination changes between camera views. In this paper we address the problem of dealing with illumination changes in order to recover matching of individuals appearing at different camera sites. This task is challenging as accurately mapping colour changes between views requires an exhaustive set of corresponding chromatic brightness values to be collected, which is very difficult in real world scenarios. We propose a Cumulative Brightness Transfer Function (CBTF) for mapping colour between cameras located at different physical sites, which makes better use of the available colour information from a very sparse training set. In addition we develop a bi-directional mapping approach to obtain a more accurate similarity measure between a pair of candidate objects. We evaluate the proposed method using challenging datasets obtained from real world distributed CCTV camera networks. The results demonstrate that our bi-directional CBTF method significantly outperforms existing techniques. 1

In this paper we address the problem of selecting the most relevant features for human identification by gait. Although gait as a behavioral biometric is concerned with how people walk rather than how people look, most existing gait recognition approaches employ both shape and dynamics information for recognition. This is because shape, as a static appearance feature also contains useful information for identification. However, the inclusion of shape information in the gait features can also introduce variations that will hinder the recognition performance. To address this problem, we develop both supervised and unsupervised feature selection methods to extract the most relevant and informative features from Gait Energy Image (GEI) for human identification. Extensive experiments are carried out which indicate that our feature selection methods significantly improve the performance of gait recognition.

This paper aims to address the problem of modelling video behaviour captured in surveillancevideos for the applications of online normal behaviour recognition and anomaly detection. A novelframework is developed for automatic behaviour profiling and online anomaly sampling/detectionwithout any manual labelling of the training dataset. The framework consists of the followingkey components: (1) A compact and effective behaviour representation method is developed basedon discrete scene event detection. The similarity between behaviour patterns are measured basedon modelling each pattern using a Dynamic Bayesian Network (DBN). (2) Natural grouping ofbehaviour patterns is discovered through a novel spectral clustering algorithm with unsupervisedmodel selection and feature selection on the eigenvectors of a normalised affinity matrix. (3) Acomposite generative behaviour model is constructed which is capable of generalising from asmall training set to accommodate variations in unseen normal behaviour patterns. (4) A run-timeaccumulative anomaly measure is introduced to detect abnormal behaviour while normal behaviourpatterns are recognised when sufficient visual evidence has become available based on an onlineLikelihood Ratio Test (LRT) method. This ensures robust and reliable anomaly detection and normalbehaviour recognition at the shortest possible time. The effectiveness and robustness of our approachis demonstrated through experiments using noisy and sparse datasets collected from both indoorand outdoor surveillance scenarios. In particular, it is shown that a behaviour model trained usingan unlabelled dataset is superior to those trained using the same but labelled dataset in detectinganomaly from an unseen video. The experiments also suggest that our online LRT based behaviourrecognition approach is advantageous over the commonly used Maximum Likelihood (ML) methodin differentiating ambiguities among different behaviour classes observed online.

We address the problem of automatic interpretation ofnon-exaggerated human facial and body behaviours captured in video. We illustrate our approach by three examples. (1) We introduce Canonical Correlation Analysis (CCA) and Matrix Canonical Correlation Analysis (MCCA) for capturing and analyzing spatial correlations among non-adjacent facial parts for facial behaviour analysis. (2) We extend Canonical Correlation Analysis to multimodality correlation for bebaviour inference using both facial and body gestures. (3) We model temporal correlation among human movement patterns in a wider space using a mixture of Multi-Observation Hidden Markov Model for human behaviour profiling and behavioural anomaly detection.

This study addresses the problem of learning the optimal structure of a dynamic graphical model for video content analysis given sparse data. We propose a Completed Likelihood AIC (CL-AIC) scoring function that differs from existing ones by optimising explicitly both the explanation and prediction capabilities of a model simultaneously. We demonstrate that CL-AIC is superior to existing scoring functions including BIC, AIC and ICL in building dynamic graph models for video content analysis. 1

We aim to tackle the problem of unsupervised visual learning. A novel relevance learning algorithm is proposed for data clustering using eigenvectors of a data affinity matrix. We show that it is critical to select the relevant eigenvectors for both estimating the optimal number of clusters and performing data clustering especially given noisy and sparse data. The effectiveness of our algorithm is demonstrated on solving two challenging visual data clustering problems: image segmentation and video behaviour profiling.

An approach is proposed for robust online behaviour recognition and abnormality detection based on discovering natural grouping of bebaviour patterns through unsupervised learning and a time accumulative reliability measure. A novel behaviour learning model and a run-time accumulative reliability measure are introduced to determine both the natural groupings of possible normal behaviour classes without manual labelling and when sufficient visual evidence has become available for differentiating ambiguities among different behaviour classes observed online. This ensures behaviour recognition at the shortest possible time and robust abnormality detection. 1

This study addresses the problem of unsupervised visual learning. It examines existing popular model order selection criteria before proposes two novel criteria for improving visual learning given sparse data and without any knowledge about model complexity. In particular, a rectified Bayesian information criterion (BICr) and a completed likelihood Akaike's information criterion (CL-AIC) are formulated to estimate the optimal model order (complexity) for learning the dynamic structure of a visual scene. Both criteria are designed to overcome poor model selection by existing popular criteria when the data sample size varies from very small to large. Extensive experiments on learning a dynamic scene structure are carried out to demonstrate the effectiveness of BICr and CL-AIC, compared to that of BIC (Schwarz, 1978), AIC (Akaike, 1973), ICL (Biernacki, 2000) and a MML (Figueiredo and Jain, 2002) based criterion.

A novel framework is developed for automatic behaviour profiling and abnormality sampling/detection without any manual labelling of the training dataset. Natural grouping of behaviour patterns is discovered through unsupervised model selection and feature selection on the eigenvectors of a normalised affinity matrix. Our experiments demonstrate that a behaviour model trained using an unlabelled dataset is superior to those trained using the same but labelled dataset in detecting abnormality from an unseen video.

A novel approach is developed to segment continuous CCTV recordings according to the activities captured in the scene. This approach differs from previous approaches which are mostly based on shot change detection and shot grouping. Video content is represented by constructing a cumulative multi-event histogram over time. An on-line segmentation algorithm is then proposed to detect breakpoints in the video content, which is more robust to noise and computationally much more efficient compared to existing on-line segmentation algorithms. Its performance is compared to that of two off-line algorithms on surveillance videos monitoring an aircraft ramp area. Our experiments demonstrate that this activity based video content representation is superior to a colour histogram based representation for surveillance video segmentation. 1

Modelling events is one of the key problems in dynamic scene understanding when salient and autonomous visual changes occurring in a scene need to be characterised as a set of different object temporal events. We propose an approach to understand complex outdoor scenarios which is based on modelling temporally correlated events using dynamic Bayesian networks (DBNs). A partially coupled hidden Markov model (PCHMM) is exploited whose topology is determined automatically using the Bayesian information criterion (BIC). Causality discovery and events modelling are also tackled using a multi-observation hidden Markov model (MOHMM).

We develop Dynamically Multi-Linked Hidden Markov Models (DML-HMMs) for interpreting group activities involving multiple objects captured in an outdoor scene. The models are based on the discovery of salient dynamic interlinks among multiple different object events. A layered hierarchical DML-HMM is built using Schwarz’s Bayesian Information Criterion (BIC) based factorisation resulting in its topology being intrinsically determined by the underlying causality and temporal order among different object events. Our experiments demonstrate that the performance of a DML-HMM on modelling group activities in a noisy outdoor scene is superior compared to that of a Coupled Hidden Markov Model (CHMM).

We put forth in this paper a geometrically motivated 3D (three-dimensional) motion error analysis, which is capable of supporting investigation of global effect such as inherent ambiguities. The error expression that we derive allows us to predict the exact conditions likely to cause ambiguities and how these ambiguities vary with motion types such as lateral or forward motion. Our formulation, though geometrically motivated, is employed to model the effect of noise.

Dynamic Probabilistic Networks (DPNs) are exploited for modeling the temporal relationships among a set of different object temporal events in the scene for a coherent and robust scene-level behaviour interpretation. In particular, we develop a Dynamically Multi-Linked Hidden Markov Model (DML-HMM) to interpret group activities involving multiple objects captured in an outdoor scene. The model is based on the discovery of salient dynamic interlinks among multiple temporal events using DPNs. Object temporal events are detected and labeled using Gaussian Mixture Models with automatic model order selection. A DML-HMM is built using Schwarz's Bayesian Information Criterion based factorisation resulting in its topology being intrinsically determined by the underlying causality and temporal order among different object events. Our experiments demonstrate that its performance on modelling group activities in a noisy outdoor scene is superior compared to that of a Multi-Observation Hidden Markov Model (MOHMM), a Parallel Hidden Markov Model (PaHMM) and a Coupled Hidden Markov Model (CHMM).

Given that most current Structure From Motion (SFM) algorithms cannot recover true motion estimates, it is important to understand the impact such motion errors have on the shape reconstruction. In this paper, we present a theoretical model to understand the impact that errors in the motion estimates have on shape recovery. Using this model, we focus on the recovery of second order shape under different generic motions, each of which presenting different degrees of error sensitivity. Understanding such different distortion behaviour is important if we want to design better fusion strategy with other shape cues. 1

Modelling events is one of the key problems in dynamic scene analysis when salient and autonomous visual changes occuring in a scene need to be characterised effectively as meaningful events. We propose a new approach for modelling such temporal events based on the local intensity temporal history of pixels. The method provides a computationally very effective temporal measure for detecting autonomous events. Events are represented and detected first at the pixel level and then at a blob level (grouped pixels) autonomously. The Expectation-Maximisation (EM) algorithm is employed to cluster events with automatic model order selection using modified Minimum Description Length (MDL). Experiments are presented to demonstrate that meaningful clusters of blob-level events can be formed without object segmentation and tracking. 1