Jiajun Liang

Shichao Dong, Jin Wang, Renhe Ji et al.

In this paper, we analyse the generalization ability of binary classifiers for the task of deepfake detection. We find that the stumbling block to their generalization is caused by the unexpected learned identity representation on images. Termed as the Implicit Identity Leakage, this phenomenon has been qualitatively and quantitatively verified among various DNNs. Furthermore, based on such understanding, we propose a simple yet effective method named the ID-unaware Deepfake Detection Model to reduce the influence of this phenomenon. Extensive experimental results demonstrate that our method outperforms the state-of-the-art in both in-dataset and cross-dataset evaluation. The code is available at https://github.com/megvii-research/CADDM.

Borui Zhao, Quan Cui, Renjie Song et al.

State-of-the-art distillation methods are mainly based on distilling deep features from intermediate layers, while the significance of logit distillation is greatly overlooked. To provide a novel viewpoint to study logit distillation, we reformulate the classical KD loss into two parts, i.e., target class knowledge distillation (TCKD) and non-target class knowledge distillation (NCKD). We empirically investigate and prove the effects of the two parts: TCKD transfers knowledge concerning the "difficulty" of training samples, while NCKD is the prominent reason why logit distillation works. More importantly, we reveal that the classical KD loss is a coupled formulation, which (1) suppresses the effectiveness of NCKD and (2) limits the flexibility to balance these two parts. To address these issues, we present Decoupled Knowledge Distillation (DKD), enabling TCKD and NCKD to play their roles more efficiently and flexibly. Compared with complex feature-based methods, our DKD achieves comparable or even better results and has better training efficiency on CIFAR-100, ImageNet, and MS-COCO datasets for image classification and object detection tasks. This paper proves the great potential of logit distillation, and we hope it will be helpful for future research. The code is available at https://github.com/megvii-research/mdistiller.

Lingfeng Yang, Xiang Li, Renjie Song et al.

Fine-grained image classification is a challenging computer vision task where various species share similar visual appearances, resulting in misclassification if merely based on visual clues. Therefore, it is helpful to leverage additional information, e.g., the locations and dates for data shooting, which can be easily accessible but rarely exploited. In this paper, we first demonstrate that existing multimodal methods fuse multiple features only on a single dimension, which essentially has insufficient help in feature discrimination. To fully explore the potential of multimodal information, we propose a dynamic MLP on top of the image representation, which interacts with multimodal features at a higher and broader dimension. The dynamic MLP is an efficient structure parameterized by the learned embeddings of variable locations and dates. It can be regarded as an adaptive nonlinear projection for generating more discriminative image representations in visual tasks. To our best knowledge, it is the first attempt to explore the idea of dynamic networks to exploit multimodal information in fine-grained image classification tasks. Extensive experiments demonstrate the effectiveness of our method. The t-SNE algorithm visually indicates that our technique improves the recognizability of image representations that are visually similar but with different categories. Furthermore, among published works across multiple fine-grained datasets, dynamic MLP consistently achieves SOTA results https://paperswithcode.com/dataset/inaturalist and takes third place in the iNaturalist challenge at FGVC8 https://www.kaggle.com/c/inaturalist-2021/leaderboard. Code is available at https://github.com/ylingfeng/DynamicMLP.git

Yuhao Chen, Xin Tan, Borui Zhao et al.

Semi-supervised learning (SSL) has attracted enormous attention due to its vast potential of mitigating the dependence on large labeled datasets. The latest methods (e.g., FixMatch) use a combination of consistency regularization and pseudo-labeling to achieve remarkable successes. However, these methods all suffer from the waste of complicated examples since all pseudo-labels have to be selected by a high threshold to filter out noisy ones. Hence, the examples with ambiguous predictions will not contribute to the training phase. For better leveraging all unlabeled examples, we propose two novel techniques: Entropy Meaning Loss (EML) and Adaptive Negative Learning (ANL). EML incorporates the prediction distribution of non-target classes into the optimization objective to avoid competition with target class, and thus generating more high-confidence predictions for selecting pseudo-label. ANL introduces the additional negative pseudo-label for all unlabeled data to leverage low-confidence examples. It adaptively allocates this label by dynamically evaluating the top-k performance of the model. EML and ANL do not introduce any additional parameter and hyperparameter. We integrate these techniques with FixMatch, and develop a simple yet powerful framework called FullMatch. Extensive experiments on several common SSL benchmarks (CIFAR-10/100, SVHN, STL-10 and ImageNet) demonstrate that FullMatch exceeds FixMatch by a large margin. Integrated with FlexMatch (an advanced FixMatch-based framework), we achieve state-of-the-art performance. Source code is at https://github.com/megvii-research/FullMatch.

Jiajun Liang, Linze Li, Zhaodong Bing et al.

Self-distillation exploits non-uniform soft supervision from itself during training and improves performance without any runtime cost. However, the overhead during training is often overlooked, and yet reducing time and memory overhead during training is increasingly important in the giant models' era. This paper proposes an efficient self-distillation method named Zipf's Label Smoothing (Zipf's LS), which uses the on-the-fly prediction of a network to generate soft supervision that conforms to Zipf distribution without using any contrastive samples or auxiliary parameters. Our idea comes from an empirical observation that when the network is duly trained the output values of a network's final softmax layer, after sorting by the magnitude and averaged across samples, should follow a distribution reminiscent to Zipf's Law in the word frequency statistics of natural languages. By enforcing this property on the sample level and throughout the whole training period, we find that the prediction accuracy can be greatly improved. Using ResNet50 on the INAT21 fine-grained classification dataset, our technique achieves +3.61% accuracy gain compared to the vanilla baseline, and 0.88% more gain against the previous label smoothing or self-distillation strategies. The implementation is publicly available at https://github.com/megvii-research/zipfls.

Siyuan Wei, Tianzhu Ye, Shen Zhang et al.

Although vision transformers (ViTs) have shown promising results in various computer vision tasks recently, their high computational cost limits their practical applications. Previous approaches that prune redundant tokens have demonstrated a good trade-off between performance and computation costs. Nevertheless, errors caused by pruning strategies can lead to significant information loss. Our quantitative experiments reveal that the impact of pruned tokens on performance should be noticeable. To address this issue, we propose a novel joint Token Pruning & Squeezing module (TPS) for compressing vision transformers with higher efficiency. Firstly, TPS adopts pruning to get the reserved and pruned subsets. Secondly, TPS squeezes the information of pruned tokens into partial reserved tokens via the unidirectional nearest-neighbor matching and similarity-based fusing steps. Compared to state-of-the-art methods, our approach outperforms them under all token pruning intensities. Especially while shrinking DeiT-tiny&small computational budgets to 35%, it improves the accuracy by 1%-6% compared with baselines on ImageNet classification. The proposed method can accelerate the throughput of DeiT-small beyond DeiT-tiny, while its accuracy surpasses DeiT-tiny by 4.78%. Experiments on various transformers demonstrate the effectiveness of our method, while analysis experiments prove our higher robustness to the errors of the token pruning policy. Code is available at https://github.com/megvii-research/TPS-CVPR2023.

Quan Cui, Bingchen Zhao, Zhao-Min Chen et al.

This work simultaneously considers the discriminability and transferability properties of deep representations in the typical supervised learning task, i.e., image classification. By a comprehensive temporal analysis, we observe a trade-off between these two properties. The discriminability keeps increasing with the training progressing while the transferability intensely diminishes in the later training period. From the perspective of information-bottleneck theory, we reveal that the incompatibility between discriminability and transferability is attributed to the over-compression of input information. More importantly, we investigate why and how the InfoNCE loss can alleviate the over-compression, and further present a learning framework, named contrastive temporal coding~(CTC), to counteract the over-compression and alleviate the incompatibility. Extensive experiments validate that CTC successfully mitigates the incompatibility, yielding discriminative and transferable representations. Noticeable improvements are achieved on the image classification task and challenging transfer learning tasks. We hope that this work will raise the significance of the transferability property in the conventional supervised learning setting. Code is available at https://github.com/DTennant/dt-tradeoff.

Ge Wu, Xin Zhang, Zheng Li et al.

Prompt learning has surfaced as an effective approach to enhance the performance of Vision-Language Models (VLMs) like CLIP when applied to downstream tasks. However, current learnable prompt tokens are primarily used for the single phase of adapting to tasks (i.e., adapting prompt), easily leading to overfitting risks. In this work, we propose a novel Cascade Prompt Learning CasPL framework to enable prompt learning to serve both generic and specific expertise (i.e., boosting and adapting prompt) simultaneously. Specifically, CasPL is a new learning paradigm comprising two distinct phases of learnable prompts: the first boosting prompt is crafted to extract domain-general knowledge from a senior larger CLIP teacher model by aligning their predicted logits using extensive unlabeled domain images. The second adapting prompt is then cascaded with the frozen first set to fine-tune the downstream tasks, following the approaches employed in prior research. In this manner, CasPL can effectively capture both domain-general and task-specific representations into explicitly different gradual groups of prompts, thus potentially alleviating overfitting issues in the target domain. It's worth noting that CasPL serves as a plug-and-play module that can seamlessly integrate into any existing prompt learning approach. CasPL achieves a significantly better balance between performance and inference speed, which is especially beneficial for deploying smaller VLM models in resource-constrained environments. Compared to the previous state-of-the-art method PromptSRC, CasPL shows an average improvement of 1.85% for base classes, 3.44% for novel classes, and 2.72% for the harmonic mean over 11 image classification datasets. Code is publicly available at: https://github.com/megvii-research/CasPL.

Shichao Dong, Jin Wang, Jiajun Liang et al.

This paper aims to interpret how deepfake detection models learn artifact features of images when just supervised by binary labels. To this end, three hypotheses from the perspective of image matching are proposed as follows. 1. Deepfake detection models indicate real/fake images based on visual concepts that are neither source-relevant nor target-relevant, that is, considering such visual concepts as artifact-relevant. 2. Besides the supervision of binary labels, deepfake detection models implicitly learn artifact-relevant visual concepts through the FST-Matching (i.e. the matching fake, source, target images) in the training set. 3. Implicitly learned artifact visual concepts through the FST-Matching in the raw training set are vulnerable to video compression. In experiments, the above hypotheses are verified among various DNNs. Furthermore, based on this understanding, we propose the FST-Matching Deepfake Detection Model to boost the performance of forgery detection on compressed videos. Experiment results show that our method achieves great performance, especially on highly-compressed (e.g. c40) videos.

Zhi Lv, Bo Lin, Siyuan Liang et al.

We present a simple domain generalization baseline, which wins second place in both the common context generalization track and the hybrid context generalization track respectively in NICO CHALLENGE 2022. We verify the founding in recent literature, domainbed, that ERM is a strong baseline compared to recent state-of-the-art domain generalization methods and propose SimpleDG which includes several simple yet effective designs that further boost generalization performance. Code is available at https://github.com/megvii-research/SimpleDG

Shuangping Jin, Bingbing Yu, Minhao Jing et al.

RGB-NIR fusion is a promising method for low-light imaging. However, high-intensity noise in low-light images amplifies the effect of structure inconsistency between RGB-NIR images, which fails existing algorithms. To handle this, we propose a new RGB-NIR fusion algorithm called Dark Vision Net (DVN) with two technical novelties: Deep Structure and Deep Inconsistency Prior (DIP). The Deep Structure extracts clear structure details in deep multiscale feature space rather than raw input space, which is more robust to noisy inputs. Based on the deep structures from both RGB and NIR domains, we introduce the DIP to leverage the structure inconsistency to guide the fusion of RGB-NIR. Benefiting from this, the proposed DVN obtains high-quality lowlight images without the visual artifacts. We also propose a new dataset called Dark Vision Dataset (DVD), consisting of aligned RGB-NIR image pairs, as the first public RGBNIR fusion benchmark. Quantitative and qualitative results on the proposed benchmark show that DVN significantly outperforms other comparison algorithms in PSNR and SSIM, especially in extremely low light conditions.

Borui Zhao, Renjie Song, Jiajun Liang

Distilling knowledge from convolutional neural networks (CNNs) is a double-edged sword for vision transformers (ViTs). It boosts the performance since the image-friendly local-inductive bias of CNN helps ViT learn faster and better, but leading to two problems: (1) Network designs of CNN and ViT are completely different, which leads to different semantic levels of intermediate features, making spatial-wise knowledge transfer methods (e.g., feature mimicking) inefficient. (2) Distilling knowledge from CNN limits the network convergence in the later training period since ViT's capability of integrating global information is suppressed by CNN's local-inductive-bias supervision. To this end, we present Cumulative Spatial Knowledge Distillation (CSKD). CSKD distills spatial-wise knowledge to all patch tokens of ViT from the corresponding spatial responses of CNN, without introducing intermediate features. Furthermore, CSKD exploits a Cumulative Knowledge Fusion (CKF) module, which introduces the global response of CNN and increasingly emphasizes its importance during the training. Applying CKF leverages CNN's local inductive bias in the early training period and gives full play to ViT's global capability in the later one. Extensive experiments and analysis on ImageNet-1k and downstream datasets demonstrate the superiority of our CSKD. Code will be publicly available.

Yuan Wang, Ouxiang Li, Yulong Xu et al.

Recent advances in generative video models are increasingly driven by post-training and test-time scaling, both of which critically depend on the quality of video reward models (RMs). An ideal reward model should predict accurate rewards that align with human preferences across diverse scenarios. However, existing paradigms face a fundamental dilemma: \textit{Discriminative RMs} regress rewards directly on features extracted by multimodal large language models (MLLMs) without explicit reasoning, making them prone to shortcut learning and heavily reliant on massive data scaling for generalization. In contrast, \textit{Generative RMs} with Chain-of-Thought (CoT) reasoning exhibit superior interpretability and generalization potential, as they leverage fine-grained semantic supervision to internalize the rationales behind human preferences. However, they suffer from inherent optimization bottlenecks due to the coupling of reasoning and scoring within a single autoregressive inference chain. To harness the generalization benefits of CoT reasoning while mitigating the training instability of coupled reasoning and scoring, we introduce DeScore, a training-efficient and generalizable video reward model. DeScore employs a decoupled ``think-then-score'' paradigm: an MLLM first generates an explicit CoT, followed by a dedicated discriminative scoring module consisting of a learnable query token and a regression head that predicts the final reward. DeScore is optimized via a two-stage framework: (1) a discriminative cold start incorporating a random mask mechanism to ensure robust scoring capabilities, and (2) a dual-objective reinforcement learning stage that independently refines CoT reasoning quality and calibrates the final reward, ensuring that higher-quality reasoning directly translates to superior model performance.

Borui Zhao, Quan Cui, Renjie Song et al.

Knowledge distillation transfers knowledge from a large model to a small one via task and distillation losses. In this paper, we observe a trade-off between task and distillation losses, i.e., introducing distillation loss limits the convergence of task loss. We believe that the trade-off results from the insufficient optimization of distillation loss. The reason is: The teacher has a lower task loss than the student, and a lower distillation loss drives the student more similar to the teacher, then a better-converged task loss could be obtained. To break the trade-off, we propose the Distillation-Oriented Trainer (DOT). DOT separately considers gradients of task and distillation losses, then applies a larger momentum to distillation loss to accelerate its optimization. We empirically prove that DOT breaks the trade-off, i.e., both losses are sufficiently optimized. Extensive experiments validate the superiority of DOT. Notably, DOT achieves a +2.59% accuracy improvement on ImageNet-1k for the ResNet50-MobileNetV1 pair. Conclusively, DOT greatly benefits the student's optimization properties in terms of loss convergence and model generalization. Code will be made publicly available.

Kling Team, Jialu Chen, Yuanzheng Ci et al.

We present Kling-Omni, a generalist generative framework designed to synthesize high-fidelity videos directly from multimodal visual language inputs. Adopting an end-to-end perspective, Kling-Omni bridges the functional separation among diverse video generation, editing, and intelligent reasoning tasks, integrating them into a holistic system. Unlike disjointed pipeline approaches, Kling-Omni supports a diverse range of user inputs, including text instructions, reference images, and video contexts, processing them into a unified multimodal representation to deliver cinematic-quality and highly-intelligent video content creation. To support these capabilities, we constructed a comprehensive data system that serves as the foundation for multimodal video creation. The framework is further empowered by efficient large-scale pre-training strategies and infrastructure optimizations for inference. Comprehensive evaluations reveal that Kling-Omni demonstrates exceptional capabilities in in-context generation, reasoning-based editing, and multimodal instruction following. Moving beyond a content creation tool, we believe Kling-Omni is a pivotal advancement toward multimodal world simulators capable of perceiving, reasoning, generating and interacting with the dynamic and complex worlds.

Shurong Yang, Huadong Li, Juhao Wu et al.

Diffusion models have demonstrated superior performance in the field of portrait animation. However, current approaches relied on either visual or audio modality to control character movements, failing to exploit the potential of mixed-modal control. This challenge arises from the difficulty in balancing the weak control strength of audio modality and the strong control strength of visual modality. To address this issue, we introduce MegActor-$Σ$: a mixed-modal conditional diffusion transformer (DiT), which can flexibly inject audio and visual modality control signals into portrait animation. Specifically, we make substantial advancements over its predecessor, MegActor, by leveraging the promising model structure of DiT and integrating audio and visual conditions through advanced modules within the DiT framework. To further achieve flexible combinations of mixed-modal control signals, we propose a ``Modality Decoupling Control" training strategy to balance the control strength between visual and audio modalities, along with the ``Amplitude Adjustment" inference strategy to freely regulate the motion amplitude of each modality. Finally, to facilitate extensive studies in this field, we design several dataset evaluation metrics to filter out public datasets and solely use this filtered dataset to train MegActor-$Σ$. Extensive experiments demonstrate the superiority of our approach in generating vivid portrait animations, outperforming previous methods trained on private dataset.

Haoran He, Yuxiao Ye, Jie Liu et al.

Fine-tuning diffusion models via online reinforcement learning (RL) has shown great potential for enhancing text-to-image alignment. However, since precisely specifying a ground-truth objective for visual tasks remains challenging, the models are often optimized using a proxy reward that only partially captures the true goal. This mismatch often leads to reward hacking, where proxy scores increase while real image quality deteriorates and generation diversity collapses. While common solutions add regularization against the reference policy to prevent reward hacking, they compromise sample efficiency and impede the exploration of novel, high-reward regions, as the reference policy is usually sub-optimal. To address the competing demands of sample efficiency, effective exploration, and mitigation of reward hacking, we propose Gated and Adaptive Regularization with Diversity-aware Optimization (GARDO), a versatile framework compatible with various RL algorithms. Our key insight is that regularization need not be applied universally; instead, it is highly effective to selectively penalize a subset of samples that exhibit high uncertainty. To address the exploration challenge, GARDO introduces an adaptive regularization mechanism wherein the reference model is periodically updated to match the capabilities of the online policy, ensuring a relevant regularization target. To address the mode collapse issue in RL, GARDO amplifies the rewards for high-quality samples that also exhibit high diversity, encouraging mode coverage without destabilizing the optimization process. Extensive experiments across diverse proxy rewards and hold-out unseen metrics consistently show that GARDO mitigates reward hacking and enhances generation diversity without sacrificing sample efficiency or exploration, highlighting its effectiveness and robustness.

Jiajun Liang, Qian Zhang, Wei Deng et al.

This work introduces a novel and efficient Bayesian federated learning algorithm, namely, the Federated Averaging stochastic Hamiltonian Monte Carlo (FA-HMC), for parameter estimation and uncertainty quantification. We establish rigorous convergence guarantees of FA-HMC on non-iid distributed data sets, under the strong convexity and Hessian smoothness assumptions. Our analysis investigates the effects of parameter space dimension, noise on gradients and momentum, and the frequency of communication (between the central node and local nodes) on the convergence and communication costs of FA-HMC. Beyond that, we establish the tightness of our analysis by showing that the convergence rate cannot be improved even for continuous FA-HMC process. Moreover, extensive empirical studies demonstrate that FA-HMC outperforms the existing Federated Averaging-Langevin Monte Carlo (FA-LD) algorithm.

Shen Zhang, Zhaowei Chen, Zhenyu Zhao et al.

Diffusion models have become a mainstream approach for high-resolution image synthesis. However, directly generating higher-resolution images from pretrained diffusion models will encounter unreasonable object duplication and exponentially increase the generation time. In this paper, we discover that object duplication arises from feature duplication in the deep blocks of the U-Net. Concurrently, We pinpoint the extended generation times to self-attention redundancy in U-Net's top blocks. To address these issues, we propose a tuning-free higher-resolution framework named HiDiffusion. Specifically, HiDiffusion contains Resolution-Aware U-Net (RAU-Net) that dynamically adjusts the feature map size to resolve object duplication and engages Modified Shifted Window Multi-head Self-Attention (MSW-MSA) that utilizes optimized window attention to reduce computations. we can integrate HiDiffusion into various pretrained diffusion models to scale image generation resolutions even to 4096x4096 at 1.5-6x the inference speed of previous methods. Extensive experiments demonstrate that our approach can address object duplication and heavy computation issues, achieving state-of-the-art performance on higher-resolution image synthesis tasks.

Yucheng Liao, Jiajun Liang, Kaiqian Cui et al.

Instruction-based image editing through natural language has emerged as a powerful paradigm for intuitive visual manipulation. While recent models achieve impressive results on single edits, they suffer from severe quality degradation under multi-turn editing. Through systematic analysis, we identify progressive loss of high-frequency information as the primary cause of this quality degradation. We present FreqEdit, a training-free framework that enables stable editing across 10+ consecutive iterations. Our approach comprises three synergistic components: (1) high-frequency feature injection from reference velocity fields to preserve fine-grained details, (2) an adaptive injection strategy that spatially modulates injection strength for precise region-specific control, and (3) a path compensation mechanism that periodically recalibrates the editing trajectory to prevent over-constraint. Extensive experiments demonstrate that FreqEdit achieves superior performance in both identity preservation and instruction following compared to seven state-of-the-art baselines.

Zhenyuan Chen, Lingfeng Yang, Shuo Chen et al.

Prompt learning is an effective method to customize Vision-Language Models (VLMs) for various downstream tasks, involving tuning very few parameters of input prompt tokens. Recently, prompt pretraining in large-scale dataset (e.g., ImageNet-21K) has played a crucial role in prompt learning for universal visual discrimination. However, we revisit and observe that the limited learnable prompts could face underfitting risks given the extensive images during prompt pretraining, simultaneously leading to poor generalization. To address the above issues, in this paper, we propose a general framework termed Revisiting Prompt Pretraining (RPP), which targets at improving the fitting and generalization ability from two aspects: prompt structure and prompt supervision. For prompt structure, we break the restriction in common practice where query, key, and value vectors are derived from the shared learnable prompt token. Instead, we introduce unshared individual query, key, and value learnable prompts, thereby enhancing the model's fitting capacity through increased parameter diversity. For prompt supervision, we additionally utilize soft labels derived from zero-shot probability predictions provided by a pretrained Contrastive Language Image Pretraining (CLIP) teacher model. These soft labels yield more nuanced and general insights into the inter-class relationships, thereby endowing the pretraining process with better generalization ability. RPP produces a more resilient prompt initialization, enhancing its robust transferability across diverse visual recognition tasks. Experiments across various benchmarks consistently confirm the state-of-the-art (SOTA) performance of our pretrained prompts. Codes and models will be made available soon.

Zhishan Zhou, Zhi Lv, Shihao Zhou et al.

This report introduce our work on Egocentric 3D Hand Pose Estimation workshop. Using AssemblyHands, this challenge focuses on egocentric 3D hand pose estimation from a single-view image. In the competition, we adopt ViT based backbones and a simple regressor for 3D keypoints prediction, which provides strong model baselines. We noticed that Hand-objects occlusions and self-occlusions lead to performance degradation, thus proposed a non-model method to merge multi-view results in the post-process stage. Moreover, We utilized test time augmentation and model ensemble to make further improvement. We also found that public dataset and rational preprocess are beneficial. Our method achieved 12.21mm MPJPE on test dataset, achieve the first place in Egocentric 3D Hand Pose Estimation challenge.

Linze Li, Zong-Wei Hong, Shen Zhang et al.

While Flow Matching theoretically guarantees constant-velocity trajectories, we identify a critical breakdown in high-dimensional practice: the Velocity Deficit. We show that the MSE objective systematically underestimates velocity magnitude, causing generated samples to fail to reach the data manifold-a phenomenon we term Integration Lag. To rectify this, we propose Initial Energy Injection, instantiated via two complementary methods: the training-based Magnitude-Aware Flow Matching (MAFM) and the training-free Scale Schedule Corrector (SSC). Both are grounded in our discovery of a crucial asymmetry: velocity contraction causes harmful kinetic stagnation at the trajectory's start, yet acts as a beneficial denoising mechanism at its end. Empirically, SSC yields significant efficiency gains with zero retraining and just one line of code. On ImageNet-1k (256x256), it improves FID by 44.6% (from 13.68 to 7.58) and achieves a 5x speedup, enabling a 50-step generator (FID 7.58) to beat a 250-step baseline (FID 8.65). Furthermore, our methods generalize to Text-to-Image tasks and high-resolution generation, improving FID on MS-COCO by ~22%.

Qunzhong Wang, Jie Liu, Jiajun Liang et al.

Recent advancements in multimodal reward models (RMs) have substantially improved post-training for visual generative models. However, current RMs face inherent limitations: (1) visual inputs consume large context budgets, forcing fewer frames and causing loss of fine-grained details; and (2) all visual information is packed into the initial prompt, exacerbating hallucination and forgetting during chain-of-thought reasoning. To overcome these issues, we introduce VideoReward Thinker (VR-Thinker), a thinking-with-image framework that equips the RM with visual reasoning operations (e.g., select frame) and a configurable visual memory window. This allows the RM to actively acquire and update visual evidence within context limits, improving reasoning fidelity and reliability. We activate visual reasoning via a reinforcement fine-tuning pipeline: (i) Cold Start with curated visual chain-of-thought data to distill basic reasoning skills and operation formatting; (ii) select samples whose per-dimension and overall judgments are all correct, then conduct Rejection sampling Fine-Tuning on these high-quality traces to further enhance reasoning; and (iii) apply Group Relative Policy Optimization (GRPO) to strengthen reasoning. Our approach delivers state-of-the-art accuracy among open-source models on video preference benchmarks, especially for longer videos: a 7B VR-Thinker achieves 80.5% on VideoGen Reward, 82.3% on GenAI-Bench, and 75.6% on MJ-Bench-Video. These results validate the effectiveness and promise of thinking-with-image multimodal reward modeling.

Jie Liu, Gongye Liu, Jiajun Liang et al.

We propose Flow-GRPO, the first method to integrate online policy gradient reinforcement learning (RL) into flow matching models. Our approach uses two key strategies: (1) an ODE-to-SDE conversion that transforms a deterministic Ordinary Differential Equation (ODE) into an equivalent Stochastic Differential Equation (SDE) that matches the original model's marginal distribution at all timesteps, enabling statistical sampling for RL exploration; and (2) a Denoising Reduction strategy that reduces training denoising steps while retaining the original number of inference steps, significantly improving sampling efficiency without sacrificing performance. Empirically, Flow-GRPO is effective across multiple text-to-image tasks. For compositional generation, RL-tuned SD3.5-M generates nearly perfect object counts, spatial relations, and fine-grained attributes, increasing GenEval accuracy from $63\%$ to $95\%$. In visual text rendering, accuracy improves from $59\%$ to $92\%$, greatly enhancing text generation. Flow-GRPO also achieves substantial gains in human preference alignment. Notably, very little reward hacking occurred, meaning rewards did not increase at the cost of appreciable image quality or diversity degradation.

Minqiang Zou, Zhi Lv, Riqiang Jin et al.

Multi-view egocentric hand tracking is a challenging task and plays a critical role in VR interaction. In this report, we present a method that uses multi-view input images and camera extrinsic parameters to estimate both hand shape and pose. To reduce overfitting to the camera layout, we apply crop jittering and extrinsic parameter noise augmentation. Additionally, we propose an offline neural smoothing post-processing method to further improve the accuracy of hand position and pose. Our method achieves 13.92mm MPJPE on the Umetrack dataset and 21.66mm MPJPE on the HOT3D dataset.

Kexin Zhu, Bo Lin, Yang Qiu et al.

We introduce a high-performance fingerprint liveness feature extraction technique that secured first place in LivDet 2023 Fingerprint Representation Challenge. Additionally, we developed a practical fingerprint recognition system with 94.68% accuracy, earning second place in LivDet 2023 Liveness Detection in Action. By investigating various methods, particularly style transfer, we demonstrate improvements in accuracy and generalization when faced with limited training data. As a result, our approach achieved state-of-the-art performance in LivDet 2023 Challenges.

Jie Liu, Gongye Liu, Jiajun Liang et al.

Video generation has achieved significant advances through rectified flow techniques, but issues like unsmooth motion and misalignment between videos and prompts persist. In this work, we develop a systematic pipeline that harnesses human feedback to mitigate these problems and refine the video generation model. Specifically, we begin by constructing a large-scale human preference dataset focused on modern video generation models, incorporating pairwise annotations across multi-dimensions. We then introduce VideoReward, a multi-dimensional video reward model, and examine how annotations and various design choices impact its rewarding efficacy. From a unified reinforcement learning perspective aimed at maximizing reward with KL regularization, we introduce three alignment algorithms for flow-based models. These include two training-time strategies: direct preference optimization for flow (Flow-DPO) and reward weighted regression for flow (Flow-RWR), and an inference-time technique, Flow-NRG, which applies reward guidance directly to noisy videos. Experimental results indicate that VideoReward significantly outperforms existing reward models, and Flow-DPO demonstrates superior performance compared to both Flow-RWR and supervised fine-tuning methods. Additionally, Flow-NRG lets users assign custom weights to multiple objectives during inference, meeting personalized video quality needs.

Zicong Fan, Takehiko Ohkawa, Linlin Yang et al.

We interact with the world with our hands and see it through our own (egocentric) perspective. A holistic 3Dunderstanding of such interactions from egocentric views is important for tasks in robotics, AR/VR, action recognition and motion generation. Accurately reconstructing such interactions in 3D is challenging due to heavy occlusion, viewpoint bias, camera distortion, and motion blur from the head movement. To this end, we designed the HANDS23 challenge based on the AssemblyHands and ARCTIC datasets with carefully designed training and testing splits. Based on the results of the top submitted methods and more recent baselines on the leaderboards, we perform a thorough analysis on 3D hand(-object) reconstruction tasks. Our analysis demonstrates the effectiveness of addressing distortion specific to egocentric cameras, adopting high-capacity transformers to learn complex hand-object interactions, and fusing predictions from different views. Our study further reveals challenging scenarios intractable with state-of-the-art methods, such as fast hand motion, object reconstruction from narrow egocentric views, and close contact between two hands and objects. Our efforts will enrich the community's knowledge foundation and facilitate future hand studies on egocentric hand-object interactions.

Zhishan Zhou, Shihao. zhou, Zhi Lv et al.

3D hand pose estimation has found broad application in areas such as gesture recognition and human-machine interaction tasks. As performance improves, the complexity of the systems also increases, which can limit the comparative analysis and practical implementation of these methods. In this paper, we propose a simple yet effective baseline that not only surpasses state-of-the-art (SOTA) methods but also demonstrates computational efficiency. To establish this baseline, we abstract existing work into two components: a token generator and a mesh regressor, and then examine their core structures. A core structure, in this context, is one that fulfills intrinsic functions, brings about significant improvements, and achieves excellent performance without unnecessary complexities. Our proposed approach is decoupled from any modifications to the backbone, making it adaptable to any modern models. Our method outperforms existing solutions, achieving state-of-the-art (SOTA) results across multiple datasets. On the FreiHAND dataset, our approach produced a PA-MPJPE of 5.7mm and a PA-MPVPE of 6.0mm. Similarly, on the Dexycb dataset, we observed a PA-MPJPE of 5.5mm and a PA-MPVPE of 5.0mm. As for performance speed, our method reached up to 33 frames per second (fps) when using HRNet and up to 70 fps when employing FastViT-MA36

Haoran He, Jiajun Liang, Xintao Wang et al.

As the marginal cost of scaling computation (data and parameters) during model pre-training continues to increase substantially, test-time scaling (TTS) has emerged as a promising direction for improving generative model performance by allocating additional computation at inference time. While TTS has demonstrated significant success across multiple language tasks, there remains a notable gap in understanding the test-time scaling behaviors of image and video generative models (diffusion-based or flow-based models). Although recent works have initiated exploration into inference-time strategies for vision tasks, these approaches face critical limitations: being constrained to task-specific domains, exhibiting poor scalability, or falling into reward over-optimization that sacrifices sample diversity. In this paper, we propose \textbf{Evo}lutionary \textbf{Search} (EvoSearch), a novel, generalist, and efficient TTS method that effectively enhances the scalability of both image and video generation across diffusion and flow models, without requiring additional training or model expansion. EvoSearch reformulates test-time scaling for diffusion and flow models as an evolutionary search problem, leveraging principles from biological evolution to efficiently explore and refine the denoising trajectory. By incorporating carefully designed selection and mutation mechanisms tailored to the stochastic differential equation denoising process, EvoSearch iteratively generates higher-quality offspring while preserving population diversity. Through extensive evaluation across both diffusion and flow architectures for image and video generation tasks, we demonstrate that our method consistently outperforms existing approaches, achieves higher diversity, and shows strong generalizability to unseen evaluation metrics. Our project is available at the website https://tinnerhrhe.github.io/evosearch.

Shen Zhang, Siyuan Liang, Yaning Tan et al.

Diffusion transformers (DiTs) struggle to generate images at resolutions higher than their training resolutions. The primary obstacle is that the explicit positional encodings(PE), such as RoPE, need extrapolating to unseen positions which degrades performance when the inference resolution differs from training. In this paper, We propose a Length-Extrapolatable Diffusion Transformer~(LEDiT) to overcome this limitation. LEDiT needs no explicit PEs, thereby avoiding PE extrapolation. The key innovation of LEDiT lies in the use of causal attention. We demonstrate that causal attention can implicitly encode global positional information and show that such information facilitates extrapolation. We further introduce a locality enhancement module, which captures fine-grained local information to complement the global coarse-grained position information encoded by causal attention. Experimental results on both conditional and text-to-image generation tasks demonstrate that LEDiT supports up to 4x resolution scaling (e.g., from 256x256 to 512x512), achieving better image quality compared to the state-of-the-art length extrapolation methods. We believe that LEDiT marks a departure from the standard RoPE-based methods and offers a promising insight into length extrapolation. Project page: https://shenzhang2145.github.io/ledit/

Zhaodong Bing, Linze Li, Jiajun Liang

Knowledge distillation (KD) in transformers often faces challenges due to misalignment in the number of attention heads between teacher and student models. Existing methods either require identical head counts or introduce projectors to bridge dimensional gaps, limiting flexibility and efficiency. We propose Squeezing-Heads Distillation (SHD), a novel approach that enables seamless knowledge transfer between models with varying head counts by compressing multi-head attention maps via efficient linear approximation. Unlike prior work, SHD eliminates alignment barriers without additional parameters or architectural modifications. Our method dynamically approximates the combined effect of multiple teacher heads into fewer student heads, preserving fine-grained attention patterns while reducing redundancy. Experiments across language (LLaMA, GPT) and vision (DiT, MDT) generative and vision (DeiT) discriminative tasks demonstrate SHD's effectiveness: it outperforms logit-based and feature-alignment KD baselines, achieving state-of-the-art results in image classification, image generation language fine-tuning, and language pre-training. The key innovations of flexible head compression, projector-free design, and linear-time complexity make SHD a versatile and scalable solution for distilling modern transformers. This work bridges a critical gap in KD, enabling efficient deployment of compact models without compromising performance.

Jing Wang, Jiajun Liang, Jie Liu et al.

Recently, GRPO-based reinforcement learning has shown remarkable progress in optimizing flow-matching models, effectively improving their alignment with task-specific rewards. Within these frameworks, the policy update relies on importance-ratio clipping to constrain overconfident positive and negative gradients. However, in practice, we observe a systematic shift in the importance-ratio distribution-its mean falls below 1 and its variance differs substantially across timesteps. This left-shifted and inconsistent distribution prevents positive-advantage samples from entering the clipped region, causing the mechanism to fail in constraining overconfident positive updates. As a result, the policy model inevitably enters an implicit over-optimization stage-while the proxy reward continues to increase, essential metrics such as image quality and text-prompt alignment deteriorate sharply, ultimately making the learned policy impractical for real-world use. To address this issue, we introduce GRPO-Guard, a simple yet effective enhancement to existing GRPO frameworks. Our method incorporates ratio normalization, which restores a balanced and step-consistent importance ratio, ensuring that PPO clipping properly constrains harmful updates across denoising timesteps. In addition, a gradient reweighting strategy equalizes policy gradients over noise conditions, preventing excessive updates from particular timestep regions. Together, these designs act as a regulated clipping mechanism, stabilizing optimization and substantially mitigating implicit over-optimization without relying on heavy KL regularization. Extensive experiments on multiple diffusion backbones (e.g., SD3.5M, Flux.1-dev) and diverse proxy tasks demonstrate that GRPO-Guard significantly reduces over-optimization while maintaining or even improving generation quality.

Zhaowei Chen, Borui Zhao, Yuchen Ge et al.

Online Knowledge Distillation (OKD) methods streamline the distillation training process into a single stage, eliminating the need for knowledge transfer from a pretrained teacher network to a more compact student network. This paper presents an innovative approach to leverage intermediate spatial representations. Our analysis of the intermediate features from both teacher and student models reveals two pivotal insights: (1) the similar features between students and teachers are predominantly focused on foreground objects. (2) teacher models emphasize foreground objects more than students. Building on these findings, we propose Asymmetric Decision-Making (ADM) to enhance feature consensus learning for student models while continuously promoting feature diversity in teacher models. Specifically, Consensus Learning for student models prioritizes spatial features with high consensus relative to teacher models. Conversely, Divergence Learning for teacher models highlights spatial features with lower similarity compared to student models, indicating superior performance by teacher models in these regions. Consequently, ADM facilitates the student models to catch up with the feature learning process of the teacher models. Extensive experiments demonstrate that ADM consistently surpasses existing OKD methods across various online knowledge distillation settings and also achieves superior results when applied to offline knowledge distillation, semantic segmentation and diffusion distillation tasks.

Jiajun Liang, Baoquan Zhang, Yunming Ye et al.

The accurate detection of Mesoscale Convective Systems (MCS) is crucial for meteorological monitoring due to their potential to cause significant destruction through severe weather phenomena such as hail, thunderstorms, and heavy rainfall. However, the existing methods for MCS detection mostly targets on single-frame detection, which just considers the static characteristics and ignores the temporal evolution in the life cycle of MCS. In this paper, we propose a novel encoder-decoder neural network for MCS detection(MCSDNet). MCSDNet has a simple architecture and is easy to expand. Different from the previous models, MCSDNet targets on multi-frames detection and leverages multi-scale spatiotemporal information for the detection of MCS regions in remote sensing imagery(RSI). As far as we know, it is the first work to utilize multi-scale spatiotemporal information to detect MCS regions. Firstly, we design a multi-scale spatiotemporal information module to extract multi-level semantic from different encoder levels, which makes our models can extract more detail spatiotemporal features. Secondly, a Spatiotemporal Mix Unit(STMU) is introduced to MCSDNet to capture both intra-frame features and inter-frame correlations, which is a scalable module and can be replaced by other spatiotemporal module, e.g., CNN, RNN, Transformer and our proposed Dual Spatiotemporal Attention(DSTA). This means that the future works about spatiotemporal modules can be easily integrated to our model. Finally, we present MCSRSI, the first publicly available dataset for multi-frames MCS detection based on visible channel images from the FY-4A satellite. We also conduct several experiments on MCSRSI and find that our proposed MCSDNet achieve the best performance on MCS detection task when comparing to other baseline methods.

Linze Li, Sunqi Fan, Hengjun Pu et al.

Over recent years, diffusion models have facilitated significant advancements in video generation. Yet, the creation of face-related videos still confronts issues such as low facial fidelity, lack of frame consistency, limited editability and uncontrollable human poses. To address these challenges, we introduce a facial animation generation method that enhances both face identity fidelity and editing capabilities while ensuring frame consistency. This approach incorporates the concept of an anchor frame to counteract the degradation of generative ability in original text-to-image models when incorporating a motion module. We propose two strategies towards this objective: training-free and training-based anchor frame methods. Our method's efficacy has been validated on multiple representative DreamBooth and LoRA models, delivering substantial improvements over the original outcomes in terms of facial fidelity, text-to-image editability, and video motion. Moreover, we introduce conditional control using a 3D parametric face model to capture accurate facial movements and expressions. This solution augments the creative possibilities for facial animation generation through the integration of multiple control signals. For additional samples, please visit https://paper-faac.github.io/.

Jiajun Liang, Chuyang Ke, Jean Honorio

In this paper, we study the information theoretic bounds for exact recovery in sub-hypergraph models for community detection. We define a general model called the $m-$uniform sub-hypergraph stochastic block model ($m-$ShSBM). Under the $m-$ShSBM, we use Fano's inequality to identify the region of model parameters where any algorithm fails to exactly recover the planted communities with a large probability. We also identify the region where a Maximum Likelihood Estimation (MLE) algorithm succeeds to exactly recover the communities with high probability. Our bounds are tight and pertain to the community detection problems in various models such as the planted hypergraph stochastic block model, the planted densest sub-hypergraph model, and the planted multipartite hypergraph model.

Minghui Liao, Jian Zhang, Zhaoyi Wan et al.

Inspired by speech recognition, recent state-of-the-art algorithms mostly consider scene text recognition as a sequence prediction problem. Though achieving excellent performance, these methods usually neglect an important fact that text in images are actually distributed in two-dimensional space. It is a nature quite different from that of speech, which is essentially a one-dimensional signal. In principle, directly compressing features of text into a one-dimensional form may lose useful information and introduce extra noise. In this paper, we approach scene text recognition from a two-dimensional perspective. A simple yet effective model, called Character Attention Fully Convolutional Network (CA-FCN), is devised for recognizing the text of arbitrary shapes. Scene text recognition is realized with a semantic segmentation network, where an attention mechanism for characters is adopted. Combined with a word formation module, CA-FCN can simultaneously recognize the script and predict the position of each character. Experiments demonstrate that the proposed algorithm outperforms previous methods on both regular and irregular text datasets. Moreover, it is proven to be more robust to imprecise localizations in the text detection phase, which are very common in practice.

Xinyu Zhou, Cong Yao, He Wen et al.

Previous approaches for scene text detection have already achieved promising performances across various benchmarks. However, they usually fall short when dealing with challenging scenarios, even when equipped with deep neural network models, because the overall performance is determined by the interplay of multiple stages and components in the pipelines. In this work, we propose a simple yet powerful pipeline that yields fast and accurate text detection in natural scenes. The pipeline directly predicts words or text lines of arbitrary orientations and quadrilateral shapes in full images, eliminating unnecessary intermediate steps (e.g., candidate aggregation and word partitioning), with a single neural network. The simplicity of our pipeline allows concentrating efforts on designing loss functions and neural network architecture. Experiments on standard datasets including ICDAR 2015, COCO-Text and MSRA-TD500 demonstrate that the proposed algorithm significantly outperforms state-of-the-art methods in terms of both accuracy and efficiency. On the ICDAR 2015 dataset, the proposed algorithm achieves an F-score of 0.7820 at 13.2fps at 720p resolution.