Junhai Yong

Haowei Zhu, Ji Liu, Ziqiong Liu et al.

Diffusion models demonstrate outstanding performance in image generation, but their multi-step inference mechanism requires immense computational cost. Previous works accelerate inference by leveraging layer or token cache techniques to reduce computational cost. However, these methods fail to achieve superior acceleration performance in few-step diffusion transformer models due to inefficient feature caching strategies, manually designed sparsity allocation, and the practice of retaining complete forward computations in several steps in these token cache methods. To tackle these challenges, we propose a differentiable layer-wise sparsity optimization framework for diffusion transformer models, leveraging token caching to reduce token computation costs and enhance acceleration. Our method optimizes layer-wise sparsity allocation in an end-to-end manner through a learnable network combined with a dynamic programming solver. Additionally, our proposed two-stage training strategy eliminates the need for full-step processing in existing methods, further improving efficiency. We conducted extensive experiments on a range of diffusion-transformer models, including DiT-XL/2, PixArt-$α$, FLUX, and Wan2.1. Across these architectures, our method consistently improves efficiency without degrading sample quality. For example, on PixArt-$α$ with 20 sampling steps, we reduce computational cost by $54\%$ while achieving generation metrics that surpass those of the original model, substantially outperforming prior approaches. These results demonstrate that our method delivers large efficiency gains while often improving generation quality.

Huanghao Yin, Shenkun Xu, Kanle Shi et al.

Text-conditioned image editing has greatly benefitted from the advancements in Image Diffusion Models. However, extending these techniques to facial video editing introduces challenges in preserving facial identity throughout the source video and ensuring consistency of the edited subject across frames. In this paper, we introduce DiffMagicFace, a unique video editing framework that integrates two fine-tuned models for text and image control. These models operate concurrently during inference to produce video frames that maintain identity features while seamlessly aligning with the editing semantics. To ensure the consistency of the edited videos, we develop a dataset comprising images showcasing various facial perspectives for each edited subject. The creation of a data set is achieved through rendering techniques and the subsequent application of optimization algorithms. Remarkably, our approach does not depend on video datasets but still delivers high-quality results in both consistency and content. The excellent effect holds even for complex tasks like talking head videos and distinguishing closely related categories. The videos edited using our framework exhibit parity with videos that are made using traditional rendering software. Through comparative analysis with current state-of-the-art methods, our framework demonstrates superior performance in both visual appeal and quantitative metrics.

Haowei Zhu, Tingxuan Huang, Xing Wang et al.

Diffusion models achieve strong generative performance but remain slow at inference due to the need for repeated full-model denoising passes. We present Token-Adaptive Predictor (TAP), a training-free, probe-driven framework that adaptively selects a predictor for each token at every sampling step. TAP uses a single full evaluation of the model's first layer as a low-cost probe to compute proxy losses for a compact family of candidate predictors (instantiated primarily with Taylor expansions of varying order and horizon), then assigns each token the predictor with the smallest proxy error. This per-token "probe-then-select" strategy exploits heterogeneous temporal dynamics, requires no additional training, and is compatible with various predictor designs. TAP incurs negligible overhead while enabling large speedups with little or no perceptual quality loss. Extensive experiments across multiple diffusion architectures and generation tasks show that TAP substantially improves the accuracy-efficiency frontier compared to fixed global predictors and caching-only baselines.

Haowei Zhu, Fangyuan Zhang, Rui Qin et al.

As the scale of vision models continues to grow, Visual Prompt Tuning (VPT) has emerged as a parameter-efficient transfer learning technique, noted for its superior performance compared to full fine-tuning. However, indiscriminately applying prompts to every layer without considering their inherent correlations, can cause significant disturbances, leading to suboptimal transferability. Additionally, VPT disrupts the original self-attention structure, affecting the aggregation of visual features, and lacks a mechanism for explicitly mining discriminative visual features, which are crucial for classification. To address these issues, we propose a Semantic Hierarchical Prompt (SHIP) fine-tuning strategy. We adaptively construct semantic hierarchies and use semantic-independent and semantic-shared prompts to learn hierarchical representations. We also integrate attribute prompts and a prompt matching loss to enhance feature discrimination and employ decoupled attention for robustness and reduced inference costs. SHIP significantly improves performance, achieving a 4.9% gain in accuracy over VPT with a ViT-B/16 backbone on VTAB-1k tasks. Our code is available at https://github.com/haoweiz23/SHIP.

Chen Ma, Shuyu Cheng, Li Chen et al.

We propose a simple and highly query-efficient black-box adversarial attack named SWITCH, which has a state-of-the-art performance in the score-based setting. SWITCH features a highly efficient and effective utilization of the gradient of a surrogate model $\hat{\mathbf{g}}$ w.r.t. the input image, i.e., the transferable gradient. In each iteration, SWITCH first tries to update the current sample along the direction of $\hat{\mathbf{g}}$, but considers switching to its opposite direction $-\hat{\mathbf{g}}$ if our algorithm detects that it does not increase the value of the attack objective function. We justify the choice of switching to the opposite direction by a local approximate linearity assumption. In SWITCH, only one or two queries are needed per iteration, but it is still effective due to the rich information provided by the transferable gradient, thereby resulting in unprecedented query efficiency. To improve the robustness of SWITCH, we further propose SWITCH$_\text{RGF}$ in which the update follows the direction of a random gradient-free (RGF) estimate when neither $\hat{\mathbf{g}}$ nor its opposite direction can increase the objective, while maintaining the advantage of SWITCH in terms of query efficiency. Experimental results conducted on CIFAR-10, CIFAR-100 and TinyImageNet show that compared with other methods, SWITCH achieves a satisfactory attack success rate using much fewer queries, and SWITCH$_\text{RGF}$ achieves the state-of-the-art attack success rate with fewer queries overall. Our approach can serve as a strong baseline for future black-box attacks because of its simplicity. The PyTorch source code is released on https://github.com/machanic/SWITCH.

Chen Ma, Li Chen, Junhai Yong

Detecting action units (AUs) on human faces is challenging because various AUs make subtle facial appearance change over various regions at different scales. Current works have attempted to recognize AUs by emphasizing important regions. However, the incorporation of expert prior knowledge into region definition remains under-exploited, and current AU detection approaches do not use regional convolutional neural networks (R-CNN) with expert prior knowledge to directly focus on AU-related regions adaptively. By incorporating expert prior knowledge, we propose a novel R-CNN based model named AU R-CNN. The proposed solution offers two main contributions: (1) AU R-CNN directly observes different facial regions, where various AUs are located. Specifically, we define an AU partition rule which encodes the expert prior knowledge into the region definition and RoI-level label definition. This design produces considerably better detection performance than existing approaches. (2) We integrate various dynamic models (including convolutional long short-term memory, two stream network, conditional random field, and temporal action localization network) into AU R-CNN and then investigate and analyze the reason behind the performance of dynamic models. Experiment results demonstrate that \textit{only} static RGB image information and no optical flow-based AU R-CNN surpasses the one fused with dynamic models. AU R-CNN is also superior to traditional CNNs that use the same backbone on varying image resolutions. State-of-the-art recognition performance of AU detection is achieved. The complete network is end-to-end trainable. Experiments on BP4D and DISFA datasets show the effectiveness of our approach. The implementation code is available online.

Yunzhe Shao, Xinyu Yi, Lu Yin et al.

This paper proposes a novel method called MagShield, designed to address the issue of magnetic interference in sparse inertial motion capture (MoCap) systems. Existing Inertial Measurement Unit (IMU) systems are prone to orientation estimation errors in magnetically disturbed environments, limiting their practical application in real-world scenarios. To address this problem, MagShield employs a "detect-then-correct" strategy, first detecting magnetic disturbances through multi-IMU joint analysis, and then correcting orientation errors using human motion priors. MagShield can be integrated with most existing sparse inertial MoCap systems, improving their performance in magnetically disturbed environments. Experimental results demonstrate that MagShield significantly enhances the accuracy of motion capture under magnetic interference and exhibits good compatibility across different sparse inertial MoCap systems.

Yumeng Zhang, Li Chen, Yufeng Liu et al.

Deep learning methods have achieved excellent performance in pose estimation, but the lack of robustness causes the keypoints to change drastically between similar images. In view of this problem, a stable heatmap regression method is proposed to alleviate network vulnerability to small perturbations. We utilize the correlation between different rows and columns in a heatmap to alleviate the multi-peaks problem, and design a highly differentiated heatmap regression to make a keypoint discriminative from surrounding points. A maximum stability training loss is used to simplify the optimization difficulty when minimizing the prediction gap of two similar images. The proposed method achieves a significant advance in robustness over state-of-the-art approaches on two benchmark datasets and maintains high performance.

Mingdong Zhang, Li Chen, Xiaoru Yuan et al.

Fencing is a sport that relies heavily on the use of tactics. However, most existing methods for analyzing fencing data are based on statistical models in which hidden patterns are difficult to discover. Unlike sequential games, such as tennis and table tennis, fencing is a type of simultaneous game. Thus, the existing methods on the sports visualization do not operate well for fencing matches. In this study, we cooperated with experts to analyze the technical and tactical characteristics of fencing competitions. To meet the requirements of the fencing experts, we designed and implemented FencingVis, an interactive visualization system for fencing competition data.The action sequences in the bout are first visualized by modified bar charts to reveal the actions of footworks and bladeworks of both fencers. Then an interactive technique is provided for exploring the patterns of behavior of fencers. The different combinations of tactical behavior patterns are further mapped to the graph model and visualized by a tactical flow graph. This graph can reveal the different strategies adopted by both fencers and their mutual influence in one bout. We also provided a number of well-coordinated views to supplement the tactical flow graph and display the information of the fencing competition from different perspectives. The well-coordinated views are meant to organically integrate with the tactical flow graph through consistent visual style and view coordination. We demonstrated the usability and effectiveness of the proposed system with three case studies. On the basis of expert feedback, FencingVis can help analysts find not only the tactical patterns hidden in fencing bouts, but also the technical and tactical characteristics of the contestant.

Yumeng Zhang, Gaoguo Jia, Li Chen et al.

There is an urgent need for an effective video classification method by means of a small number of samples. The deficiency of samples could be effectively alleviated by generating samples through Generative Adversarial Networks (GAN), but the generation of videos on a typical category remains to be underexplored since the complex actions and the changeable viewpoints are difficult to simulate. In this paper, we propose a generative data augmentation method for temporal stream of the Temporal Segment Networks with the dynamic image. The dynamic image compresses the motion information of video into a still image, removing the interference factors such as the background. Thus it is easier to generate images with categorical motion information using GAN. We use the generated dynamic images to enhance the features, with regularization achieved as well, thereby to achieve the effect of video augmentation. In order to deal with the uneven quality of generated images, we propose a Self-Paced Selection (SPS) method, which automatically selects the high-quality generated samples to be added to the network training. Our method is verified on two benchmark datasets, HMDB51 and UCF101. The experimental results show that the method can improve the accuracy of video classification under the circumstance of sample insufficiency and sample imbalance.

Yumeng Zhang, Li Chen, Yufeng Liu et al.

Insufficient labeled training datasets is one of the bottlenecks of 3D hand pose estimation from monocular RGB images. Synthetic datasets have a large number of images with precise annotations, but the obvious difference with real-world datasets impacts the generalization. Little work has been done to bridge the gap between two domains over their wide difference. In this paper, we propose a domain adaptation method called Adaptive Wasserstein Hourglass (AW Hourglass) for weakly-supervised 3D hand pose estimation, which aims to distinguish the difference and explore the common characteristics (e.g. hand structure) of synthetic and real-world datasets. Learning the common characteristics helps the network focus on pose-related information. The similarity of the characteristics makes it easier to enforce domain-invariant constraints. During training, based on the relation between these common characteristics and 3D pose learned from fully-annotated synthetic datasets, it is beneficial for the network to restore the 3D pose of weakly labeled real-world datasets with the aid of 2D annotations and depth images. While in testing, the network predicts the 3D pose with the input of RGB.

Chen Ma, Chenxu Zhao, Hailin Shi et al.

Deep neural networks (DNNs) are vulnerable to adversarial attack which is maliciously implemented by adding human-imperceptible perturbation to images and thus leads to incorrect prediction. Existing studies have proposed various methods to detect the new adversarial attacks. However, new attack methods keep evolving constantly and yield new adversarial examples to bypass the existing detectors. It needs to collect tens of thousands samples to train detectors, while the new attacks evolve much more frequently than the high-cost data collection. Thus, this situation leads the newly evolved attack samples to remain in small scales. To solve such few-shot problem with the evolving attack, we propose a meta-learning based robust detection method to detect new adversarial attacks with limited examples. Specifically, the learning consists of a double-network framework: a task-dedicated network and a master network which alternatively learn the detection capability for either seen attack or a new attack. To validate the effectiveness of our approach, we construct the benchmarks with few-shot-fashion protocols based on three conventional datasets, i.e. CIFAR-10, MNIST and Fashion-MNIST. Comprehensive experiments are conducted on them to verify the superiority of our approach with respect to the traditional adversarial attack detection methods.

Zihao Bo, Hao Zhang, Junhai Yong et al.

We propose a real-time DNN-based technique to segment hand and object of interacting motions from depth inputs. Our model is called DenseAttentionSeg, which contains a dense attention mechanism to fuse information in different scales and improves the results quality with skip-connections. Besides, we introduce a contour loss in model training, which helps to generate accurate hand and object boundaries. Finally, we propose and release our InterSegHands dataset, a fine-scale hand segmentation dataset containing about 52k depth maps of hand-object interactions. Our experiments evaluate the effectiveness of our techniques and datasets, and indicate that our method outperforms the current state-of-the-art deep segmentation methods on interaction segmentation.