Animating virtual characters with holistic co-speech gestures is a challenging but critical task. Previous systems have primarily focused on the weak correlation between audio and gestures, leading to physically unnatural outcomes that degrade the user experience. To address this problem, we introduce HoleGest, a novel neural network framework based on decoupled diffusion and motion priors for the automatic generation of high-quality, expressive co-speech gestures. Our system leverages large-scale human motion datasets to learn a robust prior with low audio dependency and high motion reliance, enabling stable global motion and detailed finger movements. To improve the generation efficiency of diffusion-based models, we integrate implicit joint constraints with explicit geometric and conditional constraints, capturing complex motion distributions between large strides. This integration significantly enhances generation speed while maintaining high-quality motion. Furthermore, we design a shared embedding spacec for gesture-transcription text alignment, enabling the generation of semantically correct gesture actions. Extensive experiments and user feedback demonstrate the effectiveness and potential applications of our model, with our method achieving a level of realism close to the ground truth, providing an immersive user experience.

Diffusion models have demonstrated remarkable synthesis quality and diversity in generating co-speech gestures. However, the computationally intensive sampling steps associated with diffusion models hinder their practicality in real-world applications. Hence, we present DIDiffGes, for a Decoupled Semi-Implicit Diffusion model-based framework, that can synthesize high-quality, expressive gestures from speech using only a few sampling steps. Our approach leverages Generative Adversarial Networks (GANs) to enable large-step sampling for diffusion model. We decouple gesture data into body and hands distributions and further decompose them into marginal and conditional distributions. GANs model the marginal distribution implicitly, while L2 reconstruction loss learns the conditional distributions exciplictly. This strategy enhances GAN training stability and ensures expressiveness of generated full-body gestures. Our framework also learns to denoise root noise conditioned on local body representation, guaranteeing stability and realism. DIDiffGes can generate gestures from speech with just 10 sampling steps, without compromising quality and expressiveness, reducing the number of sampling steps by a factor of 100 compared to existing methods. Our user study reveals that our method outperforms state-of-the-art approaches in human likeness, appropriateness, and style correctness.

Human-object interaction (HOI) synthesis is crucial for creating immersive and realistic experiences for applications such as virtual reality. Existing methods often rely on simplified object representations, such as the object’s centroid or the nearest point to a human, to achieve physically plausible motions. However, these approaches may overlook geometric complexity, resulting in suboptimal interaction fidelity. To address this limitation, we introduce ROG, a novel diffusion-based framework that models the spatiotemporal relationships inherent in HOIs with rich geometric detail. For efficient object representation, we select boundary-focused and fine-detail key points from the object mesh, ensuring a comprehensive depiction of the object’s geometry. This representation is used to construct an interactive distance field (IDF), capturing the robust HOI dynamics. Furthermore, we develop a diffusion-based relation model that integrates spatial and temporal attention mechanisms, enabling a better understanding of intricate HOI relationships. This relation model refines the generated motion’s IDF, guiding the motion generation process to produce relation-aware and semantically aligned movements. Experimental evaluations demonstrate that ROG significantly outperforms state-of-the-art methods in the realism and semantic accuracy of synthesized HOIs. This paper’s code will be released at https://lalalfhdh.github.io/rogpage/.

Open vocabulary Human-Object Interaction (HOI) detection is a challenging task that detects all triplets of interest in an image, even those that are not pre-defined in the training set. Existing approaches typically rely on output features generated by large Vision-Language Models (VLMs) to enhance the generalization ability of interaction representations. However, the visual features produced by VLMs are holistic and coarse-grained, which contradicts the nature of detection tasks. To address this issue, we propose a novel Bilateral Collaboration framework for open vocabulary HOI detection (BC-HOI). This framework includes an Attention Bias Guidance (ABG) component, which guides the VLM to produce fine-grained instance-level interaction features according to the attention bias provided by the HOI detector. It also includes a Large Language Model (LLM)-based Supervision Guidance (LSG) component, which provides fine-grained token-level supervision for the HOI detector by the LLM component of the VLM. LSG enhances the ability of ABG to generate high-quality attention bias. We conduct extensive experiments on two popular benchmarks: HICO-DET and V-COCO, consistently achieving superior performance in the open vocabulary and closed settings. The code will be released in Github.

Co-speech gesture video generation aims to synthesize realistic, audio-aligned videos of speakers, complete with synchronized facial expressions and body gestures. This task presents challenges due to the significant one-to-many mapping between audio and visual content, further complicated by the scarcity of large-scale public datasets and high computational demands. We propose a lightweight framework that utilizes 2D full-body skeletons as an efficient auxiliary condition to bridge audio signals with visual outputs. Our approach introduces a diffusion model conditioned on fine-grained audio segments and a skeleton extracted from the speaker's reference image, predicting skeletal motions through skeleton-audio feature fusion to ensure strict audio coordination and body shape consistency. The generated skeletons are then fed into an off-the-shelf human video generation model with the speaker's reference image to synthesize high-fidelity videos. To democratize research, we present CSG-405-the first public dataset with 405 hours of high-resolution videos across 71 speech types, annotated with 2D skeletons and diverse speaker demographics. Experiments show that our method exceeds state-of-the-art approaches in visual quality and synchronization while generalizing across speakers and contexts. Code, models, and CSG-405 are publicly released at https://mpi-lab.github.io/Democratizing-CSG/

Audio-driven simultaneous gesture generation is vital for human-computer communication, AI games, and film pro-duction. While previous research has shown promise, there are still limitations. Methods based on VAEs are accompa-nied by issues of local jitter and global instability, whereas methods based on diffusion models are hampered by low generation efficiency. This is because the denoising process of DDPM in the latter relies on the assumption that the noise added at each step is sampled from a unimodal distribution, and the noise values are small. DDIM bor-rows the idea from the Euler method for solving differential equations, disrupts the Markov chain process, and increases the noise step size to reduce the number of denoising steps, thereby accelerating generation. However, simply increasing the step size during the step-by-step denoising process causes the results to gradually deviate from the original data distribution, leading to a significant drop in the quality of the generated actions and the emergence of unnatural artifacts. In this paper, we break the assumptions of DDPM and achieves breakthrough progress in denoising speed and fidelity. Specifically, we introduce a conditional GAN to capture audio control signals and implicitly match the multimodal denoising distribution between the diffusion and denoising steps within the same sampling step, aiming to sample larger noise values and apply fewer denoising steps for high-speed generation. In addition, to enable the model to generate high-fidelity global gestures and avoid artifacts, we introduce an explicit motion geometric loss to enhance the quality and global stability of the generated gestures. Numerous qualitative and quantitative experiments show that compared to contemporary diffusion-based methods, our method offers faster generation speed and higher fidelity, and compared to non-diffusion methods, it provides a more stable global effect and a more natural user experience.

Generating human motion from textual descriptions is a challenging task. Existing methods either struggle with physical credibility or are limited by the complexities of physics simulations. In this paper, we present ReinDiffuse that combines reinforcement learning with motion diffusion model to generate physically credible human motions that align with textual descriptions. Our method adapts Motion Diffusion Model to output a parameterized distribution of actions, making them compatible with reinforcement learning paradigms. We employ reinforcement learning with the objective of maximizing physically plausible rewards to optimize motion generation for physical fidelity. Our approach outperforms existing state-of-the-art models on two major datasets, HumanML3D and KIT-ML, achieving significant improvements in physical plausibility and motion quality. Project: https://reindiffuse.github.io/

This supplementary material offers additional experimental results and in-depth explanations to augment the content of the main paper. In Section 1, we elaborate extensively on the preliminary work. Subsequently, in Section 2, we delve into the implementation details of the Rope module. Following this, in Section 3, we provide a wealth of experimental data and model parameters pertaining to the effectiveness of the Rope module, facilitating a more intuitive comparison. Furthermore, we supply a comprehensive visualization of the attention results and corresponding analysis. Lastly, we present a plethora of challenging data to compare and validate the efficacy of our method in handling real-world occlusion issues, and discuss its limitations and instances of failure. Beyond this, we further elucidate the details of TrajDenoiser and provide prospects for applications in 3D space.

We introduce HuTuMotion, an innovative approach for generating natural human motions that navigates latent motion diffusion models by leveraging few-shot human feedback. Unlike existing approaches that sample latent variables from a standard normal prior distribution, our method adapts the prior distribution to better suit the characteristics of the data, as indicated by human feedback, thus enhancing the quality of motion generation. Furthermore, our findings reveal that utilizing few-shot feedback can yield performance levels on par with those attained through extensive human feedback. This discovery emphasizes the potential and efficiency of incorporating few-shot human-guided optimization within latent diffusion models for personalized and style-aware human motion generation applications. The experimental results show the significantly superior performance of our method over existing state-of-the-art approaches.

Character animation in real-world scenarios necessitates a variety of constraints, such as trajectories, keyframes, interactions, etc. Existing methodologies typically treat single or a finite set of these constraint(s) as separate control tasks. These methods are often specialized, and the tasks they address are rarely extendable or customizable. We categorize these as solutions to the close-set motion control problem. In response to the complexity of practical motion control, we propose and attempt to solve the open-set motion control problem. This problem is characterized by an open and fully customizable set of motion control tasks. To address this, we introduce a new paradigm, programmable motion generation. In this paradigm, any given motion control task is broken down into a combination of atomic constraints. These constraints are then programmed into an error function that quantifies the degree to which a motion sequence adheres to them. We utilize a pretrained motion generation model and optimize its latent code to minimize the error function of the generated motion. Consequently, the generated motion not only inherits the prior of the generative model but also satisfies the requirements of the compounded constraints. Our experiments demonstrate that our approach can generate high-quality motions when addressing a wide range of unseen tasks. These tasks encompass motion control by motion dynamics, geometric constraints, physical laws, interactions with scenes, objects or the character's own body parts, etc. All of these are achieved in a unified approach, without the need for ad-hoc paired training data collection or specialized network designs. During the programming of novel tasks, we observed the emergence of new skills beyond those of the prior model. With the assistance of large language models, we also achieved automatic programming. We hope that this work will pave the way for the motion control of general AI agents.

Current talking avatars mostly generate co-speech gestures based on audio and text of the utterance, without considering the non-speaking motion of the speaker. Furthermore, previous works on co-speech gesture generation have designed network structures based on individual gesture datasets, which results in limited data volume, compromised generalizability, and restricted speaker movements. To tackle these issues, we introduce FreeTalker, which, to the best of our knowledge, is the first framework for the generation of both spontaneous (e.g., co-speech gesture) and non-spontaneous (e.g., moving around the podium) speaker motions. Specifically, we train a diffusion-based model for speaker motion generation that employs unified representations of both speech-driven gestures and text-driven motions, utilizing heterogeneous data sourced from various motion datasets. During inference, we utilize classifier-free guidance to highly control the style in the clips. Additionally, to create smooth transitions between clips, we utilize DoubleTake, a method that leverages a generative prior and ensures seamless motion blending. Extensive experiments show that our method generates natural and controllable speaker movements. Our code, model, and demo are are available at https://youngseng.github.io/FreeTalker/.

Previous motion generation methods are limited to the pre-rigged 3D human model, hindering their applications in the animation of various non-rigged characters. In this work, we present TapMo, a Text-driven Animation PIpeline for synthesizing Motion in a broad spectrum of skeleton-free 3D characters. The pivotal innovation in TapMo is its use of shape deformation-aware features as a condition to guide the diffusion model, thereby enabling the generation of mesh-specific motions for various characters. Specifically, TapMo comprises two main components - Mesh Handle Predictor and Shape-aware Diffusion Module. Mesh Handle Predictor predicts the skinning weights and clusters mesh vertices into adaptive handles for deformation control, which eliminates the need for traditional skeletal rigging. Shape-aware Motion Diffusion synthesizes motion with mesh-specific adaptations. This module employs text-guided motions and mesh features extracted during the first stage, preserving the geometric integrity of the animations by accounting for the character's shape and deformation. Trained in a weakly-supervised manner, TapMo can accommodate a multitude of non-human meshes, both with and without associated text motions. We demonstrate the effectiveness and generalizability of TapMo through rigorous qualitative and quantitative experiments. Our results reveal that TapMo consistently outperforms existing auto-animation methods, delivering superior-quality animations for both seen or unseen heterogeneous 3D characters.

Existing gesture generation methods primarily focus on upper body gestures based on audio features, neglecting speech content, emotion, and locomotion. These limitations result in stiff, mechanical gestures that fail to convey the true meaning of audio content. We introduce ExpGest, a novel framework leveraging synchronized text and audio information to generate expressive full-body gestures. Unlike AdaIN or one-hot encoding methods, we design a noise emotion classifier for optimizing adversarial direction noise, avoiding melody distortion and guiding results towards specified emotions. Moreover, aligning semantic and gestures in the latent space provides better generalization capabilities. ExpGest, a diffusion model-based gesture generation framework, is the first attempt to offer mixed generation modes, including audio-driven gestures and text-shaped motion. Experiments show that our framework effectively learns from combined text-driven motion and audio-induced gesture datasets, and preliminary results demonstrate that ExpGest achieves more expressive, natural, and controllable global motion in speakers compared to state-of-the-art models.

We present a simple yet effective method for skeleton-agnostic motion retargeting. Previous methods transfer motion between high-resolution meshes, failing to preserve the inherent local-part motions in the mesh. Addressing this issue, our proposed method learns the correspondence in a coarse-to-fine fashion by disentangling the retargeting process within multi-scale meshes. First, we propose a mesh-pooling module that pools the mesh representations for better motion transfer. This module improves the ability to handle small-part motion and preserves the local motion interdependence between neighboring mesh vertices. Furthermore, we leverage a multi-scale refinement procedure to complement missing mesh details by gradually refining the low-resolution mesh output with a higher-resolution one. We evaluate our method on several well-known 3D character datasets, and it yields an average improvement of 25% on point-wise mesh Euclidean distance (PMD) against the start-of-art method. Qualitative results show that our method is significantly helpful in preserving the moving consistency of different body parts on the target character due to disentangling body-part structures and mesh details in a multi-scale way.

The state space models, employing recursively propagated features, demonstrate strong representation capabilities comparable to Transformer models and superior efficiency. However, constrained by the inherent geometric constraints of sequences, it still falls short in modeling long-range dependencies. To address this issue, we propose the MambaTree network, which first dynamically generates a tree topology based on spatial relationships and input features. Then, feature propagation is performed based on this graph, thereby breaking the original sequence constraints to achieve stronger representation capabilities. Additionally, we introduce a linear complexity dynamic programming algorithm to enhance long-range interactions without increasing computational cost. MambaTree is a versatile multimodal framework that can be applied to both visual and textual tasks. Extensive experiments demonstrate that our method significantly outperforms existing structured state space models on image classification, object detection and segmentation. Besides, by fine-tuning large language models, our approach achieves consistent improvements in multiple textual tasks at minor training cost.

Abstract Deep models trained by using clean data have achieved tremendous success in fine-grained image classification. Yet, they generally suffer from significant performance degradation when encountering noisy labels. Existing approaches to handle label noise, though proved to be effective for generic object recognition, usually fail on fine-grained data. The reason is that, on fine-grained data, the category difference is subtle and the training sample size is small. Then deep models could easily overfit the noisy labels. To improve the robustness of deep models on noisy data for fine-grained visual categorization, in this paper, we propose a novel learning framework named ProtoSimi. Our method employs an adaptive label correction strategy, ensuring effective learning on limited data. Specifically, our approach considers the criteria of exploring the effectiveness of both global class-prototype and part class-prototype similarities in identifying and correcting labels of samples. We evaluate our method on three standard benchmarks of fine-grained recognition. Experimental results show that our method outperforms the existing label noisy methods by a large margin. In ablation studies, we also verify that our method is non-sensitive to hyper-parameters selection and can be integrated with other FGVC methods to increase the generalization performance.

Joint hand and object pose estimation from a single image is extremely challenging as serious occlusion often occurs when the hand and object interact. Existing approaches typically first extract coarse hand and object features from a single backbone, then further enhance them with reference to each other via interaction modules. However, these works usually ignore that the hand and object are competitive in feature learning, since the backbone takes both of them as foreground and they are usually mutually occluded. In this paper, we propose a novel Harmonious Feature Learning Network (HFL-Net). HFL-Net introduces a new framework that combines the advantages of single-and double-stream backbones: it shares the parameters of the low-and high-level convolutional layers of a common ResNet-50 model for the hand and object, leaving the middle-level layers unshared. This strategy enables the hand and the object to be extracted as the sole targets by the middle-level layers, avoiding their competition in feature learning. The shared high-level layers also force their features to be harmonious, thereby facilitating their mutual feature enhancement. In particular, we propose to enhance the feature of the hand via concatenation with the feature in the same location from the object stream. A subsequent self-attention layer is adopted to deeply fuse the concatenated feature. Experimental results show that our proposed approach consistently outperforms state-of-the-art methods on the popular HO3D and Dex-Ycb databases. Notably, the performance of our model on hand pose estimation even surpasses that of existing works that only perform the single-hand pose estimation task. Code is available at https://github.com/lzjJf12/HFL-Net.

Transformer-based models achieve favorable performance in artistic style transfer recently thanks to its global receptive field and powerful multi-head/layer attention operations. Nevertheless, the over-paramerized multi-layer structure increases parameters significantly and thus presents a heavy burden for training. Moreover, for the task of style transfer, vanilla Transformer that fuses content and style features by residual connections is prone to content-wise distortion. In this paper, we devise a novel Transformer model termed as Master specifically for style transfer. On the one hand, in the proposed model, different Transformer layers share a common group of parameters, which (1) reduces the total number of parameters, (2) leads to more robust training convergence, and (3) is readily to control the degree of stylization via tuning the number of stacked layers freely during inference. On the other hand, different from the vanilla version, we adopt a learnable scaling operation on content features before content-style feature interaction, which better preserves the original similarity between a pair of content features while ensuring the stylization quality. We also propose a novel meta learning scheme for the proposed model so that it can not only work in the typical setting of arbitrary style transfer, but also adaptable to the few-shot setting, by only fine-tuning the Transformer encoder layer in the few-shot stage for one specific style. Text-guided few-shot style transfer is firstly achieved with the proposed framework. Extensive experiments demonstrate the superiority of Master under both zero-shot and few-shot style transfer settings.

Reconstructing two hands from monocular RGB images is challenging due to frequent occlusion and mutual confusion. Existing methods mainly learn an entangled representation to encode two interacting hands, which are in-credibly fragile to impaired interaction, such as truncated hands, separate hands, or external occlusion. This paper presents ACR (Attention Collaboration-based Regres-sor), which makes the first attempt to reconstruct hands in arbitrary scenarios. To achieve this, ACR explicitly miti-gates interdependencies between hands and between parts by leveraging center and part-based attention for feature extraction. However, reducing interdependence helps re-lease the input constraint while weakening the mutual reasoning about reconstructing the interacting hands. Thus, based on center attention, ACR also learns cross-hand prior that handle the interacting hands better. We evaluate our method on various types of hand reconstruction datasets. Our method significantly outperforms the best interacting-hand approaches on the InterHand2.6M dataset while yielding comparable performance with the state-of-the-art single-hand methods on the FreiHand dataset. More qualitative results on in-the-wild and hand-object interaction datasets and web images/videos further demonstrate the effectiveness of our approach for arbitrary hand reconstruction. Our code is available at this link11https://github.com/ZhengdiYu/Arbitrary-Hands-3D-Reconstruction.

A good motion retargeting cannot be reached without reasonable consideration of source-target differences on both the skeleton and shape geometry levels. In this work, we propose a novel Residual RETargeting network (R2ET) structure, which relies on two neural modification modules, to adjust the source motions to fit the target skeletons and shapes progressively. In particular, a skeleton-aware module is introduced to preserve the source motion semantics. A shape-aware module is designed to perceive the geometries of target characters to reduce interpenetration and contact-missing. Driven by our explored distance-based losses that explicitly model the motion semantics and geometry, these two modules can learn residual motion modifications on the source motion to generate plausible retargeted motion in a single inference without postprocessing. To balance these two modifications, we further present a balancing gate to conduct linear interpolation between them. Extensive experiments on the public dataset Mixamo demonstrate that our R2ET achieves the state-of-the-art performance, and provides a good balance between the preservation of motion semantics as well as the attenuation of interpenetration and contact-missing. Code is available at https://github.com/Kebii/R2ET.

This paper proposes an anchor-based deformation model, namely AnchorDEF, to predict 3D garment animation from a body motion sequence. It deforms a garment mesh template by a mixture of rigid transformations with extra nonlinear displacements. A set of anchors around the mesh surface is introduced to guide the learning of rigid transformation matrices. Once the anchor transformations are found, per-vertex nonlinear displacements of the garment template can be regressed in a canonical space, which reduces the complexity of deformation space learning. By explicitly constraining the transformed anchors to satisfy the consistencies of position, normal and direction, the physical meaning of learned anchor transformations in space is guaranteed for better generalization. Furthermore, an adaptive anchor updating is proposed to optimize the anchor position by being aware of local mesh topology for learning representative anchor transformations. Qualitative and quantitative experiments on different types of garments demonstrate that AnchorDEF achieves the state-of-the-art performance on 3D garment deformation prediction in motion, especially for loose-fitting garments.

Understanding and analyzing animal behavior is increasingly essential to protect endangered animal species. However, the application of advanced computer vision techniques in this regard is minimal, which boils down to lacking large and diverse datasets for training deep models. To break the deadlock, we present LoTE-Animal, a large-scale endangered animal dataset collected over 12 years, to foster the application of deep learning in rare species conservation. The collected data contains vast variations such as ecological seasons, weather conditions, periods, viewpoints, and habitat scenes. So far, we retrieved at least 500K videos and 1.2 million images. Specifically, we selected and annotated 11 endangered animals for behavior understanding, including 10K video sequences for the action recognition task, 28K images for object detection, instance segmentation, and pose estimation tasks. In addition, we gathered 7K web images of the same species as source domain data for the domain adaptation task. We provide evaluation results of representative vision understanding approaches and cross-domain experiments. LoTE-Animal dataset would facilitate the community to research more advanced machine learning models and explore new tasks to aid endangered animal conservation. Our dataset will be released with the paper. Our dataset can be found at https://LoTE-Animal.github.io

Gestures are non-verbal but important behaviors accompanying people’s speech. While previous methods are able to generate speech rhythm-synchronized gestures, the semantic context of the speech is generally lacking in the gesticulations. Although semantic gestures do not occur very regularly in human speech, they are indeed the key for the audience to understand the speech context in a more immersive environment. Hence, we introduce LivelySpeaker, a framework that realizes semantics-aware co-speech gesture generation and offers several control handles. In particular, our method decouples the task into two stages: script-based gesture generation and audio-guided rhythm refinement. Specifically, the script-based gesture generation leverages the pre-trained CLIP text embeddings as the guidance for generating gestures that are highly semantically aligned with the script. Then, we devise a simple but effective diffusion-based gesture generation backbone simply using pure MLPs, that is conditioned on only audio signals and learns to gesticulate with realistic motions. We utilize such powerful prior to rhyme the script-guided gestures with the audio signals, notably in a zero-shot setting. Our novel two-stage generation framework also enables several applications, such as changing the gesticulation style, editing the co-speech gestures via textual prompting, and controlling the semantic awareness and rhythm alignment with guided diffusion. Extensive experiments demonstrate the advantages of the proposed framework over competing methods. In addition, our core diffusion-based generative model also achieves state-of-the-art performance on two benchmarks. The code and model will be released to facilitate future research.

Preserving geometric structures in the scene plays a vital role in image stitching. However, most of the existing methods ignore the large-scale layouts reflected by straight lines or curves, decreasing overall stitching quality. To address this issue, this work presents a structure-preserving stitching approach that produces images with natural visual effects and less distortion. Our method first employs deep learning-based edge detection to extract various types of large-scale edges. Then, the extracted edges are sampled to construct multiple groups of triangles to represent geometric structures. Meanwhile, a GEometric Structure preserving (GES) energy term is introduced to make these triangles undergo similarity transformation. Further, an optimized GES energy term is presented to reasonably determine the weights of the sampling points on the geometric structure, and the term is added into the Global Similarity Prior (GSP) stitching model called GES-GSP to achieve a smooth transition between local alignment and geometric structure preservation. The effectiveness of GES-GSP is validated through comprehensive experiments on a stitching dataset. The experimental results show that the proposed method outperforms several state-of-the-art methods in geometric structure preservation and obtains more natural stitching results. The code and dataset are available at https://github.com/flowerDuo/GES-GSP-Stitching.

Differentiating subcategories of a common visual category is challenging because of the similar appearance shared among different classes in fine-grained recognition. Existing mixture-of-expert based methods divide the fine-grained space into some specific regions and solve the integrated problem by conquering subspace ones. However, it is not feasible to learn diverse experts directly through data partition strategy because of limited data available for fine-grained recognition problems. To address the issue, we leverage visual attention to learn an enhanced experts' mixture. Specifically, we introduce a gradually-enhanced learning strategy from model attention. The strategy promotes diversity among experts by feeding each expert with full-size data distinct in granularity. We further promote expert's learning by providing it with a larger data space, which is achieved by swapping attentive regions within positive pairs. Our method learns new experts on the dataset with the prior knowledge from former experts sequentially and enforces the experts to learn more diverse but discriminative representation. These enhanced experts are finally combined to make stronger predictions. We conduct extensive experiments on fine-grained benchmarks. The results show that our method consistently outperforms the state-of-the-art method in both weakly supervised localization and fine-grained image classification. Our code is publicly available at https://github.com/lbzhang/Enhanced-Expert-FGVC-Pytorch.git.

Data mixing augmentation has proved effective in training deep models. Recent methods mix labels mainly according to the mixture proportion of image pixels. Due to the major discriminative information of a fine-grained image usually resides in subtle regions, these methods tend to introduce heavy label noise in fine-grained recognition. We propose Semantically Proportional Mixing (SnapMix) that exploits class activation map (CAM) to lessen the label noise in augmenting fine-grained data. SnapMix generates the target label for a mixed image by estimating its intrinsic semantic composition. This strategy can adapt to asymmetric mixing operations and ensure semantic correspondence between synthetic images and target labels. Experiments show that our method consistently outperforms existing mixed-based approaches regardless of different datasets or network depths. Further, by incorporating the mid-level features, the proposed SnapMix achieves top-level performance, demonstrating its potential to serve as a strong baseline for fine-grained recognition.

Learning mid-level representation for fine-grained recognition is easily dominated by a limited number of highly discriminative patterns, degrading its robustness and generalization capability. To this end, we propose a novel Stochastic Partial Swap (SPS)1scheme to address this issue. Our method performs element-wise swapping for partial features between samples to inject noise during training. It equips a regularization effect similar to Dropout, which promotes more neurons to represent the concepts. Furthermore, it also exhibits other advantages: 1) suppressing over-activation to some part patterns to improve feature representativeness, and 2) enriching pattern combination and simulating noisy cases to enhance classifier generalization. We verify the effectiveness of our approach through comprehensive experiments across four network backbones and three fine-grained datasets. Moreover, we demonstrate its ability to complement high-level representations, allowing a simple model to achieve performance comparable to the top-performing technologies in fine-grained recognition, indoor scene recognition, and material recognition while improving model interpretability.

We consider the person image synthesis problem, where an output image is generated from an arbitrary source image and an arbitrary target 3D pose. Prior person image synthesizing methods usually use 2D keypoint heatmaps to represent the target pose. However, this 2D representation can be ambiguous in self-occlusion scenarios due to the lack of depth information, resulting in generating inappropriate images. To solve this problem, we propose to synthesize person image from 3D poses. We introduce an improved part affinity field representation to describe the 3D configuration of the target pose and the 2D location of the person in the pixel space. Compared to using 2D poses, our synthesized images have visually better details, such as correct self-occlusion, brightness, face direction et al. Moreover, in contrast with prior person image generators, our method predicts the difference between the source image and target image instead of the output images. This strategy allows the network to generate better foreground and significantly reduces the noise in the image background. We evaluate our method on images of fifteen different actions on Human3.6M dataset. Extensive experiments demonstrate that our method can synthesize much better person images than 2D pose-based ones. If given a sequence of desired poses, our method can produce a sequence of temporally smooth and coherent images, even on another subject and actions, which means that our method has great potential to generate high-quality person videos.

There has been an increasing emphasis on building large-scale datasets as the driver of deep learning-based trackers' success. However, accurately annotating tracking data is highly labor-intensive and expensive, making it infeasible in real-world applications. In this study, we investigate the necessity of large-scale training data to ensure tracking algorithms' performance. To this end, we introduce a FAT (Few-Annotation Tracking) benchmark constructed by sampling one or a few frames per video from some existing tracking datasets. The proposed dataset can be used to evaluate the effectiveness of tracking algorithms considering data efficiency and new data augmentation approaches for object tracking. We further present AMMC (Augmentation by Mimicking Motion Change), a data augmentation strategy that enables learning high-performing trackers using small-scale datasets. AMMC first cuts out the tracked targets and performs a sequence of transformations to simulate the possible change by object motion. Then the transformed targets are pasted on the inpainted background images and further conjointly augmented to mimic variability caused by camera motion. Compared with standard augmentation methods, AMMC explicitly considers tracking data characteristics, which synthesizes more valid data for object tracking. We extensively evaluate our approach with two popular trackers on the FAT datasets. Experiments show that our method allows these trackers to even trained on a dataset requiring much less annotation to achieve comparable or even better performance to those on the full-annotation dataset. The results imply complete video annotation might not be necessary for object tracking if leveraging motion-driven data augmentations during training.

Recent works such as Mixup and CutMix have demonstrated the effectiveness of augmenting training data for deep models. These methods generate new data by generally blending random image contents and mixing their labels proportionally. However, this strategy tends to produce unreasonable training samples for fine-grained recognition, leading to limited improvement. This is because mixing random image contents may potentially produce images containing destructed object structures. Further, as the category differences mainly reside in small part regions, mixing labels proportionally to the number of mixed pixels might result in label noisy problem. To augment more reasonable training data, we propose Intra-class Part Swapping (InPS) that produces new data by performing attention-guided content swapping on input pairs from the same class. Compared with previous approaches, InPS avoids introducing noisy labels and ensures a likely holistic structure of objects in generated images. We demonstrate InPS outperforms the most recent augmentation approaches in both fine-grained recognition and weakly object localization. Further, by simply incorporating the mid-level feature learning, our proposed method achieves state-of-the-art performance in the literature while maintaining the simplicity and inference efficiency. Our code is publicly available.

Fine-grained visual categorization aims to learn a robust image representation modeling subtle differences from similar categories. Existing methods in this field tackle the problem by designing complex frameworks, which produce high-level features by performing first-order or second-order pooling. Despite the impressive performance achieved by these strategies, the single-order networks only carry linear or non-linear information of the last convolutional layer, neglecting the fact that features from different orders are mutually complementary. In this paper, we propose a multi-order feature statistical method (MOFS), which learns fine-grained features characterizing multiple orders. Specifically, the MOFS consists of two sub-modules: (i) a first-order module modeling both mid-level and high-level features. (ii) a covariance feature statistical module capturing high-order features. By deploying these two sub-modules on the top of existing backbone networks, MOFS simultaneously captures multi-level of discriminative patters including local, global and co-related patters. We evaluate the proposed method on three challenging benchmarks, namely CUB-200-2011, Stanford Cars, and FGVC-Aircraft. Compared with state-of-the-art methods, experiment results exhibit superior performance in recognizing fine-grained objects.

Not all the human body parts have the same degree of freedom (DOF) due to the physiological structure. For example, the limbs may move more flexibly and freely than the torso does. Most of the existing 3D pose estimation methods, despite the very promising results achieved, treat the body joints equally and consequently often lead to larger reconstruction errors on the limbs. In this paper, we propose a progressive approach that explicitly accounts for the distinct DOFs among the body parts. We model parts with higher DOFs like the elbows, as dependent components of the corresponding parts with lower DOFs like the torso, of which the 3D locations can be more reliably estimated. Meanwhile, the high-DOF parts may, in turn, impose a constraint on where the low-DOF ones lie. As a result, parts with different DOFs supervise one another, yielding physically constrained and plausible pose-estimation results. To further facilitate the prediction of the high-DOF parts, we introduce a pose-attribution estimation, where the relative location of a limb joint with respect to the torso, which has the least DOF of a human body, is explicitly estimated and further fed to the joint-estimation module. The proposed approach achieves very promising results, outperforming the state of the art on several benchmarks.

We aim to divide the problem space of fine-grained recognition into some specific regions. To achieve this, we develop a unified framework based on a mixture of experts. Due to limited data available for the fine-grained recognition problem, it is not feasible to learn diverse experts by using a data division strategy. To tackle the problem, we promote diversity among experts by combing an expert gradually-enhanced learning strategy and a Kullback-Leibler divergence based constraint. The strategy learns new experts on the dataset with the prior knowledge from former experts and adds them to the model sequentially, while the introduced constraint forces the experts to produce diverse prediction distribution. These drive the experts to learn the task from different aspects, making them specialized in different subspace problems. Experiments show that the resulting model improves the classification performance and achieves the state-of-the-art performance on several fine-grained benchmark datasets.

Learning visual representations from web data has recently attracted attention for object recognition. Previous studies have mainly focused on overcoming label noise and data bias and have shown promising results by learning directly from web data. However, we argue that it might be better to transfer knowledge from existing human labeling resources to improve performance at nearly no additional cost. In this paper, we propose a new semi-supervised method for learning via web data. Our method has the unique design of exploiting strong supervision, i.e., in addition to standard image-level labels, our method also utilizes detailed annotations including object bounding boxes and part landmarks. By transferring as much knowledge as possible from existing strongly supervised datasets to weakly supervised web images, our method can benefit from sophisticated object recognition algorithms and overcome several typical problems found in webly-supervised learning. We consider the problem of fine-grained visual categorization, in which existing training resources are scarce, as our main research objective. Comprehensive experimentation and extensive analysis demonstrate encouraging performance of the proposed approach, which, at the same time, delivers a new pipeline for fine-grained visual categorization that is likely to be highly effective for real-world applications.

We propose a coarse-fine network (CFN) that exploits multi-level supervisions for keypoint localization. Recently, convolutional neural networks (CNNs)-based methods have achieved great success due to the powerful hierarchical features in CNNs. These methods typically use confidence maps generated from ground-truth keypoint locations as supervisory signals. However, while some keypoints can be easily located with high accuracy, many of them are hard to localize due to appearance ambiguity. Thus, using strict supervision often fails to detect keypoints that are difficult to locate accurately To target this problem, we develop a keypoint localization network composed of several coarse detector branches, each of which is built on top of a feature layer in a CNN, and a fine detector branch built on top of multiple feature layers. We supervise each branch by a specified label map to explicate a certain supervision strictness level. All the branches are unified principally to produce the final accurate keypoint locations. We demonstrate the efficacy, efficiency, and generality of our method on several benchmarks for multiple tasks including bird part localization and human body pose estimation. Especially, our method achieves 72.2% AP on the 2016 COCO Keypoints Challenge dataset, which is an 18% improvement over the winning entry.

Multi-instance learning (MIL) is widely acknowledged as a fundamental method to solve weakly supervised problems. While MIL is usually effective in standard weakly supervised object recognition tasks, in this paper, we investigate the applicability of MIL on an extreme case of weakly supervised learning on the task of fine-grained visual categorization, in which intra-class variance could be larger than inter-class due to the subtle differences between subordinate categories. For this challenging task, we propose a new method that generalizes the standard multi-instance learning framework, for which a novel multi-task co-localization algorithm is proposed to take advantage of the relationship among fine-grained categories and meanwhile performs as an effective initialization strategy for the non-convex multi-instance objective. The localization results also enable object-level domain-specific fine-tuning of deep neural networks, which significantly boosts the performance. Experimental results on three fine-grained datasets reveal the effectiveness of the proposed method, especially the importance of exploiting inter-class relationships between object categories in weakly supervised fine-grained recognition.

In the context of fine-grained visual categorization, the ability to interpret models as human-understandable visual manuals is sometimes as important as achieving high classification accuracy. In this paper, we propose a novel Part-Stacked CNN architecture that explicitly explains the finegrained recognition process by modeling subtle differences from object parts. Based on manually-labeled strong part annotations, the proposed architecture consists of a fully convolutional network to locate multiple object parts and a two-stream classification network that encodes object-level and part-level cues simultaneously. By adopting a set of sharing strategies between the computation of multiple object parts, the proposed architecture is very efficient running at 20 frames/sec during inference. Experimental results on the CUB-200-2011 dataset reveal the effectiveness of the proposed architecture, from multiple perspectives of classification accuracy, model interpretability, and efficiency. Being able to provide interpretable recognition results in realtime, the proposed method is believed to be effective in practical applications.

We propose a new method for fine-grained object recognition that employs part-level annotations and deep convolutional neural networks (CNNs) in a unified framework. Although both schemes have been widely used to boost recognition performance, due to the difficulty in acquiring detailed part annotations, strongly supervised fine-grained datasets are usually too small to keep pace with the rapid evolution of CNN architectures. In this paper, we solve this problem by exploiting inexhaustible web data. The proposed method improves classification accuracy in two ways: more discriminative CNN feature representations are generated using a training set augmented by collecting a large number of part patches from weakly supervised web images, and more robust object classifiers are learned using a multi-instance learning algorithm jointly on the strong and weak datasets. Despite its simplicity, the proposed method delivers a remarkable performance improvement on the CUB200-2011 dataset compared to baseline part-based R-CNN methods, and achieves the highest accuracy on this dataset even in the absence of test image annotations.