Yichen Wei

Peiyu Wang, Yichen Wei, Yi Peng et al.

We present Skywork R1V2, a next-generation multimodal reasoning model and a major leap forward from its predecessor, Skywork R1V. At its core, R1V2 introduces a hybrid reinforcement learning paradigm that jointly leverages the Mixed Preference Optimization (MPO) and the Group Relative Policy Optimization (GRPO), which harmonizes reward-model guidance with rule-based strategies, thereby addressing the long-standing challenge of balancing sophisticated reasoning capabilities with broad generalization. To further enhance training efficiency, we propose the Selective Sample Buffer (SSB) mechanism, which effectively addresses the vanishing advantages dilemma inherent in GRPO by prioritizing high-value samples throughout the optimization process. Notably, we observe that excessive reinforcement signals can induce visual hallucinations--a phenomenon we systematically monitor and mitigate through calibrated reward thresholds throughout the training process. Empirical results affirm the exceptional capability of R1V2, with benchmark-leading performances such as 62.6 on OlympiadBench, 78.9 on AIME2024, 63.6 on LiveCodeBench, and 73.6 on MMMU. These results underscore R1V2's superiority over existing open-source models and demonstrate significant progress in closing the performance gap with premier proprietary systems, including Gemini 2.5 and OpenAI-o4-mini. The Skywork R1V2 model weights have been publicly released to promote openness and reproducibility https://huggingface.co/Skywork/Skywork-R1V2-38B.

Bin Dong, Fangao Zeng, Tiancai Wang et al.

In this paper, we propose an end-to-end framework for instance segmentation. Based on the recently introduced DETR [1], our method, termed SOLQ, segments objects by learning unified queries. In SOLQ, each query represents one object and has multiple representations: class, location and mask. The object queries learned perform classification, box regression and mask encoding simultaneously in an unified vector form. During training phase, the mask vectors encoded are supervised by the compression coding of raw spatial masks. In inference time, mask vectors produced can be directly transformed to spatial masks by the inverse process of compression coding. Experimental results show that SOLQ can achieve state-of-the-art performance, surpassing most of existing approaches. Moreover, the joint learning of unified query representation can greatly improve the detection performance of DETR. We hope our SOLQ can serve as a strong baseline for the Transformer-based instance segmentation. Code is available at https://github.com/megvii-research/SOLQ.

Fangao Zeng, Bin Dong, Yuang Zhang et al.

Temporal modeling of objects is a key challenge in multiple object tracking (MOT). Existing methods track by associating detections through motion-based and appearance-based similarity heuristics. The post-processing nature of association prevents end-to-end exploitation of temporal variations in video sequence. In this paper, we propose MOTR, which extends DETR and introduces track query to model the tracked instances in the entire video. Track query is transferred and updated frame-by-frame to perform iterative prediction over time. We propose tracklet-aware label assignment to train track queries and newborn object queries. We further propose temporal aggregation network and collective average loss to enhance temporal relation modeling. Experimental results on DanceTrack show that MOTR significantly outperforms state-of-the-art method, ByteTrack by 6.5% on HOTA metric. On MOT17, MOTR outperforms our concurrent works, TrackFormer and TransTrack, on association performance. MOTR can serve as a stronger baseline for future research on temporal modeling and Transformer-based trackers. Code is available at https://github.com/megvii-research/MOTR.

Yifan Sun, Yuke Zhu, Yuhan Zhang et al.

This paper introduces a new fundamental characteristic, \ie, the dynamic range, from real-world metric tools to deep visual recognition. In metrology, the dynamic range is a basic quality of a metric tool, indicating its flexibility to accommodate various scales. Larger dynamic range offers higher flexibility. In visual recognition, the multiple scale problem also exist. Different visual concepts may have different semantic scales. For example, ``Animal'' and ``Plants'' have a large semantic scale while ``Elk'' has a much smaller one. Under a small semantic scale, two different elks may look quite \emph{different} to each other . However, under a large semantic scale (\eg, animals and plants), these two elks should be measured as being \emph{similar}. %We argue that such flexibility is also important for deep metric learning, because different visual concepts indeed correspond to different semantic scales. Introducing the dynamic range to deep metric learning, we get a novel computer vision task, \ie, the Dynamic Metric Learning. It aims to learn a scalable metric space to accommodate visual concepts across multiple semantic scales. Based on three types of images, \emph{i.e.}, vehicle, animal and online products, we construct three datasets for Dynamic Metric Learning. We benchmark these datasets with popular deep metric learning methods and find Dynamic Metric Learning to be very challenging. The major difficulty lies in a conflict between different scales: the discriminative ability under a small scale usually compromises the discriminative ability under a large one, and vice versa. As a minor contribution, we propose Cross-Scale Learning (CSL) to alleviate such conflict. We show that CSL consistently improves the baseline on all the three datasets. The datasets and the code will be publicly available at https://github.com/SupetZYK/DynamicMetricLearning.

Cheng Zou, Bohan Wang, Yue Hu et al.

We propose HOI Transformer to tackle human object interaction (HOI) detection in an end-to-end manner. Current approaches either decouple HOI task into separated stages of object detection and interaction classification or introduce surrogate interaction problem. In contrast, our method, named HOI Transformer, streamlines the HOI pipeline by eliminating the need for many hand-designed components. HOI Transformer reasons about the relations of objects and humans from global image context and directly predicts HOI instances in parallel. A quintuple matching loss is introduced to force HOI predictions in a unified way. Our method is conceptually much simpler and demonstrates improved accuracy. Without bells and whistles, HOI Transformer achieves $26.61\% $ $ AP $ on HICO-DET and $52.9\%$ $AP_{role}$ on V-COCO, surpassing previous methods with the advantage of being much simpler. We hope our approach will serve as a simple and effective alternative for HOI tasks. Code is available at https://github.com/bbepoch/HoiTransformer .

Junjie Yan, Ruosi Wan, Xiangyu Zhang et al.

Batch Normalization (BN) is one of the most widely used techniques in Deep Learning field. But its performance can awfully degrade with insufficient batch size. This weakness limits the usage of BN on many computer vision tasks like detection or segmentation, where batch size is usually small due to the constraint of memory consumption. Therefore many modified normalization techniques have been proposed, which either fail to restore the performance of BN completely, or have to introduce additional nonlinear operations in inference procedure and increase huge consumption. In this paper, we reveal that there are two extra batch statistics involved in backward propagation of BN, on which has never been well discussed before. The extra batch statistics associated with gradients also can severely affect the training of deep neural network. Based on our analysis, we propose a novel normalization method, named Moving Average Batch Normalization (MABN). MABN can completely restore the performance of vanilla BN in small batch cases, without introducing any additional nonlinear operations in inference procedure. We prove the benefits of MABN by both theoretical analysis and experiments. Our experiments demonstrate the effectiveness of MABN in multiple computer vision tasks including ImageNet and COCO. The code has been released in https://github.com/megvii-model/MABN.

Bin Xiao, Haiping Wu, Yichen Wei

There has been significant progress on pose estimation and increasing interests on pose tracking in recent years. At the same time, the overall algorithm and system complexity increases as well, making the algorithm analysis and comparison more difficult. This work provides simple and effective baseline methods. They are helpful for inspiring and evaluating new ideas for the field. State-of-the-art results are achieved on challenging benchmarks. The code will be available at https://github.com/leoxiaobin/pose.pytorch.

Yuke Zhu, Yan Bai, Yichen Wei

Data augmentation in feature space is effective to increase data diversity. Previous methods assume that different classes have the same covariance in their feature distributions. Thus, feature transform between different classes is performed via translation. However, this approach is no longer valid for recent deep metric learning scenarios, where feature normalization is widely adopted and all features lie on a hypersphere. This work proposes a novel spherical feature transform approach. It relaxes the assumption of identical covariance between classes to an assumption of similar covariances of different classes on a hypersphere. Consequently, the feature transform is performed by a rotation that respects the spherical data distributions. We provide a simple and effective training method, and in depth analysis on the relation between the two different transforms. Comprehensive experiments on various deep metric learning benchmarks and different baselines verify that our method achieves consistent performance improvement and state-of-the-art results.

Youcai Zhang, Zhonghao Lan, Yuchen Dai et al.

Knowledge distillation(KD) aims to improve the performance of a student network by mimicing the knowledge from a powerful teacher network. Existing methods focus on studying what knowledge should be transferred and treat all samples equally during training. This paper introduces the adaptive sample weighting to KD. We discover that previous effective hard mining methods are not appropriate for distillation. Furthermore, we propose Prime-Aware Adaptive Distillation (PAD) by the incorporation of uncertainty learning. PAD perceives the prime samples in distillation and then emphasizes their effect adaptively. PAD is fundamentally different from and would refine existing methods with the innovative view of unequal training. For this reason, PAD is versatile and has been applied in various tasks including classification, metric learning, and object detection. With ten teacher-student combinations on six datasets, PAD promotes the performance of existing distillation methods and outperforms recent state-of-the-art methods.

Zechun Liu, Xiangyu Zhang, Zhiqiang Shen et al.

We present joint multi-dimension pruning (abbreviated as JointPruning), an effective method of pruning a network on three crucial aspects: spatial, depth and channel simultaneously. To tackle these three naturally different dimensions, we proposed a general framework by defining pruning as seeking the best pruning vector (i.e., the numerical value of layer-wise channel number, spacial size, depth) and construct a unique mapping from the pruning vector to the pruned network structures. Then we optimize the pruning vector with gradient update and model joint pruning as a numerical gradient optimization process. To overcome the challenge that there is no explicit function between the loss and the pruning vectors, we proposed self-adapted stochastic gradient estimation to construct a gradient path through network loss to pruning vectors and enable efficient gradient update. We show that the joint strategy discovers a better status than previous studies that focused on individual dimensions solely, as our method is optimized collaboratively across the three dimensions in a single end-to-end training and it is more efficient than the previous exhaustive methods. Extensive experiments on large-scale ImageNet dataset across a variety of network architectures MobileNet V1&V2&V3 and ResNet demonstrate the effectiveness of our proposed method. For instance, we achieve significant margins of 2.5% and 2.6% improvement over the state-of-the-art approach on the already compact MobileNet V1&V2 under an extremely large compression ratio.

Jie Chang, Zhonghao Lan, Changmao Cheng et al.

Modeling data uncertainty is important for noisy images, but seldom explored for face recognition. The pioneer work, PFE, considers uncertainty by modeling each face image embedding as a Gaussian distribution. It is quite effective. However, it uses fixed feature (mean of the Gaussian) from an existing model. It only estimates the variance and relies on an ad-hoc and costly metric. Thus, it is not easy to use. It is unclear how uncertainty affects feature learning. This work applies data uncertainty learning to face recognition, such that the feature (mean) and uncertainty (variance) are learnt simultaneously, for the first time. Two learning methods are proposed. They are easy to use and outperform existing deterministic methods as well as PFE on challenging unconstrained scenarios. We also provide insightful analysis on how incorporating uncertainty estimation helps reducing the adverse effects of noisy samples and affects the feature learning.

Huawei Wei, Peng Lu, Yichen Wei

Face alignment is crucial for face recognition and has been widely adopted. However, current practice is too simple and under-explored. There lacks an understanding of how important face alignment is and how it should be performed, for recognition. This work studies these problems and makes two contributions. First, it provides an in-depth and quantitative study of how alignment strength affects recognition accuracy. Our results show that excessive alignment is harmful and an optimal balanced point of alignment is in need. To strike the balance, our second contribution is a novel joint learning approach where alignment learning is controllable with respect to its strength and driven by recognition. Our proposed method is validated by comprehensive experiments on several benchmarks, especially the challenging ones with large pose.

Yifan Sun, Changmao Cheng, Yuhan Zhang et al.

This paper provides a pair similarity optimization viewpoint on deep feature learning, aiming to maximize the within-class similarity $s_p$ and minimize the between-class similarity $s_n$. We find a majority of loss functions, including the triplet loss and the softmax plus cross-entropy loss, embed $s_n$ and $s_p$ into similarity pairs and seek to reduce $(s_n-s_p)$. Such an optimization manner is inflexible, because the penalty strength on every single similarity score is restricted to be equal. Our intuition is that if a similarity score deviates far from the optimum, it should be emphasized. To this end, we simply re-weight each similarity to highlight the less-optimized similarity scores. It results in a Circle loss, which is named due to its circular decision boundary. The Circle loss has a unified formula for two elemental deep feature learning approaches, i.e. learning with class-level labels and pair-wise labels. Analytically, we show that the Circle loss offers a more flexible optimization approach towards a more definite convergence target, compared with the loss functions optimizing $(s_n-s_p)$. Experimentally, we demonstrate the superiority of the Circle loss on a variety of deep feature learning tasks. On face recognition, person re-identification, as well as several fine-grained image retrieval datasets, the achieved performance is on par with the state of the art.

Ruihang Chu, Yifan Sun, Yadong Li et al.

This paper considers vehicle re-identification (re-ID) problem. The extreme viewpoint variation (up to 180 degrees) poses great challenges for existing approaches. Inspired by the behavior in human's recognition process, we propose a novel viewpoint-aware metric learning approach. It learns two metrics for similar viewpoints and different viewpoints in two feature spaces, respectively, giving rise to viewpoint-aware network (VANet). During training, two types of constraints are applied jointly. During inference, viewpoint is firstly estimated and the corresponding metric is used. Experimental results confirm that VANet significantly improves re-ID accuracy, especially when the pair is observed from different viewpoints. Our method establishes the new state-of-the-art on two benchmarks.

Zichao Guo, Xiangyu Zhang, Haoyuan Mu et al.

We revisit the one-shot Neural Architecture Search (NAS) paradigm and analyze its advantages over existing NAS approaches. Existing one-shot method, however, is hard to train and not yet effective on large scale datasets like ImageNet. This work propose a Single Path One-Shot model to address the challenge in the training. Our central idea is to construct a simplified supernet, where all architectures are single paths so that weight co-adaption problem is alleviated. Training is performed by uniform path sampling. All architectures (and their weights) are trained fully and equally. Comprehensive experiments verify that our approach is flexible and effective. It is easy to train and fast to search. It effortlessly supports complex search spaces (e.g., building blocks, channel, mixed-precision quantization) and different search constraints (e.g., FLOPs, latency). It is thus convenient to use for various needs. It achieves start-of-the-art performance on the large dataset ImageNet.

Wenbo Li, Zhicheng Wang, Binyi Yin et al.

Existing pose estimation approaches fall into two categories: single-stage and multi-stage methods. While multi-stage methods are seemingly more suited for the task, their performance in current practice is not as good as single-stage methods. This work studies this issue. We argue that the current multi-stage methods' unsatisfactory performance comes from the insufficiency in various design choices. We propose several improvements, including the single-stage module design, cross stage feature aggregation, and coarse-to-fine supervision. The resulting method establishes the new state-of-the-art on both MS COCO and MPII Human Pose dataset, justifying the effectiveness of a multi-stage architecture. The source code is publicly available for further research.

Jiayuan Gu, Han Hu, Liwei Wang et al.

While most steps in the modern object detection methods are learnable, the region feature extraction step remains largely hand-crafted, featured by RoI pooling methods. This work proposes a general viewpoint that unifies existing region feature extraction methods and a novel method that is end-to-end learnable. The proposed method removes most heuristic choices and outperforms its RoI pooling counterparts. It moves further towards fully learnable object detection.

Xiangyun Zhao, Shuang Liang, Yichen Wei

In this work, we present a novel and effective framework to facilitate object detection with the instance-level segmentation information that is only supervised by bounding box annotation. Starting from the joint object detection and instance segmentation network, we propose to recursively estimate the pseudo ground-truth object masks from the instance-level object segmentation network training, and then enhance the detection network with top-down segmentation feedbacks. The pseudo ground truth mask and network parameters are optimized alternatively to mutually benefit each other. To obtain the promising pseudo masks in each iteration, we embed a graphical inference that incorporates the low-level image appearance consistency and the bounding box annotations to refine the segmentation masks predicted by the segmentation network. Our approach progressively improves the object detection performance by incorporating the detailed pixel-wise information learned from the weakly-supervised segmentation network. Extensive evaluation on the detection task in PASCAL VOC 2007 and 2012 [12] verifies that the proposed approach is effective.

Xizhou Zhu, Jifeng Dai, Lu Yuan et al.

There has been significant progresses for image object detection in recent years. Nevertheless, video object detection has received little attention, although it is more challenging and more important in practical scenarios. Built upon the recent works, this work proposes a unified approach based on the principle of multi-frame end-to-end learning of features and cross-frame motion. Our approach extends prior works with three new techniques and steadily pushes forward the performance envelope (speed-accuracy tradeoff), towards high performance video object detection.

Han Hu, Jiayuan Gu, Zheng Zhang et al.

Although it is well believed for years that modeling relations between objects would help object recognition, there has not been evidence that the idea is working in the deep learning era. All state-of-the-art object detection systems still rely on recognizing object instances individually, without exploiting their relations during learning. This work proposes an object relation module. It processes a set of objects simultaneously through interaction between their appearance feature and geometry, thus allowing modeling of their relations. It is lightweight and in-place. It does not require additional supervision and is easy to embed in existing networks. It is shown effective on improving object recognition and duplicate removal steps in the modern object detection pipeline. It verifies the efficacy of modeling object relations in CNN based detection. It gives rise to the first fully end-to-end object detector.

Xiao Sun, Bin Xiao, Fangyin Wei et al.

State-of-the-art human pose estimation methods are based on heat map representation. In spite of the good performance, the representation has a few issues in nature, such as not differentiable and quantization error. This work shows that a simple integral operation relates and unifies the heat map representation and joint regression, thus avoiding the above issues. It is differentiable, efficient, and compatible with any heat map based methods. Its effectiveness is convincingly validated via comprehensive ablation experiments under various settings, specifically on 3D pose estimation, for the first time.

Xingyi Zhou, Qixing Huang, Xiao Sun et al.

In this paper, we study the task of 3D human pose estimation in the wild. This task is challenging due to lack of training data, as existing datasets are either in the wild images with 2D pose or in the lab images with 3D pose. We propose a weakly-supervised transfer learning method that uses mixed 2D and 3D labels in a unified deep neutral network that presents two-stage cascaded structure. Our network augments a state-of-the-art 2D pose estimation sub-network with a 3D depth regression sub-network. Unlike previous two stage approaches that train the two sub-networks sequentially and separately, our training is end-to-end and fully exploits the correlation between the 2D pose and depth estimation sub-tasks. The deep features are better learnt through shared representations. In doing so, the 3D pose labels in controlled lab environments are transferred to in the wild images. In addition, we introduce a 3D geometric constraint to regularize the 3D pose prediction, which is effective in the absence of ground truth depth labels. Our method achieves competitive results on both 2D and 3D benchmarks.

Xiao Sun, Jiaxiang Shang, Shuang Liang et al.

Regression based methods are not performing as well as detection based methods for human pose estimation. A central problem is that the structural information in the pose is not well exploited in the previous regression methods. In this work, we propose a structure-aware regression approach. It adopts a reparameterized pose representation using bones instead of joints. It exploits the joint connection structure to define a compositional loss function that encodes the long range interactions in the pose. It is simple, effective, and general for both 2D and 3D pose estimation in a unified setting. Comprehensive evaluation validates the effectiveness of our approach. It significantly advances the state-of-the-art on Human3.6M and is competitive with state-of-the-art results on MPII.

Xizhou Zhu, Yuwen Xiong, Jifeng Dai et al.

Deep convolutional neutral networks have achieved great success on image recognition tasks. Yet, it is non-trivial to transfer the state-of-the-art image recognition networks to videos as per-frame evaluation is too slow and unaffordable. We present deep feature flow, a fast and accurate framework for video recognition. It runs the expensive convolutional sub-network only on sparse key frames and propagates their deep feature maps to other frames via a flow field. It achieves significant speedup as flow computation is relatively fast. The end-to-end training of the whole architecture significantly boosts the recognition accuracy. Deep feature flow is flexible and general. It is validated on two recent large scale video datasets. It makes a large step towards practical video recognition.

Xingyi Zhou, Xiao Sun, Wei Zhang et al.

Learning articulated object pose is inherently difficult because the pose is high dimensional but has many structural constraints. Most existing work do not model such constraints and does not guarantee the geometric validity of their pose estimation, therefore requiring a post-processing to recover the correct geometry if desired, which is cumbersome and sub-optimal. In this work, we propose to directly embed a kinematic object model into the deep neutral network learning for general articulated object pose estimation. The kinematic function is defined on the appropriately parameterized object motion variables. It is differentiable and can be used in the gradient descent based optimization in network training. The prior knowledge on the object geometric model is fully exploited and the structure is guaranteed to be valid. We show convincing experiment results on a toy example and the 3D human pose estimation problem. For the latter we achieve state-of-the-art result on Human3.6M dataset.

Luwei Yang, Ligen Zhu, Yichen Wei et al.

Previous part-based attribute recognition approaches perform part detection and attribute recognition in separate steps. The parts are not optimized for attribute recognition and therefore could be sub-optimal. We present an end-to-end deep learning approach to overcome the limitation. It generates object parts from key points and perform attribute recognition accordingly, allowing adaptive spatial transform of the parts. Both key point estimation and attribute recognition are learnt jointly in a multi-task setting. Extensive experiments on two datasets verify the efficacy of proposed end-to-end approach.

Xingyi Zhou, Qingfu Wan, Wei Zhang et al.

Previous learning based hand pose estimation methods does not fully exploit the prior information in hand model geometry. Instead, they usually rely a separate model fitting step to generate valid hand poses. Such a post processing is inconvenient and sub-optimal. In this work, we propose a model based deep learning approach that adopts a forward kinematics based layer to ensure the geometric validity of estimated poses. For the first time, we show that embedding such a non-linear generative process in deep learning is feasible for hand pose estimation. Our approach is verified on challenging public datasets and achieves state-of-the-art performance.