A5026239594- Adversarial Robustness in Machine Learning
- Video Surveillance and Tracking Methods
- Anomaly Detection Techniques and Applications
- Advanced Image Processing Techniques
- Advanced Neural Network Applications
- Digital Media Forensic Detection
- Image Enhancement Techniques
- Generative Adversarial Networks and Image Synthesis
- Bacillus and Francisella bacterial research
- Advanced Vision and Imaging
- Image and Signal Denoising Methods
- Wireless Communication Networks Research
- Advanced Image and Video Retrieval Techniques
- Advanced Image Fusion Techniques
- Satellite Communication Systems
- Speech Recognition and Synthesis
- Face recognition and analysis
- Natural Language Processing Techniques
- Visual Attention and Saliency Detection
- Advanced Computational Techniques and Applications
- Integrated Circuits and Semiconductor Failure Analysis
- Recommender Systems and Techniques
- Human Pose and Action Recognition
- Infrared Target Detection Methodologies
- Fire Detection and Safety Systems
Institute of High Performance Computing
2023-2025
Agency for Science, Technology and Research
2023-2025
Harbin Institute of Technology
2003-2025
China Three Gorges University
2011-2024
Nanyang Technological University
2015-2023
China Energy Engineering Corporation (China)
2023
Air Force Engineering University
2016-2023
Guilin University of Aerospace Technology
2023
Beijing University of Chemical Technology
2015-2023
Chinese Academy of Cultural Heritage
2023
How to effectively learn temporal variation of target appearance, exclude the interference cluttered background, while maintaining real-time response, is an essential problem visual object tracking. Recently, Siamese networks have shown great potentials matching based trackers in achieving balanced accuracy and beyond realtime speed. However, they still a big gap classification & updating tolerating changes objects imaging conditions. In this paper, we propose dynamic network, via fast...
As the GAN-based face image and video generation techniques, widely known as DeepFakes, have become more matured realistic, there comes a pressing urgent demand for effective DeepFakes detectors. Motivated by fact that remote visual photoplethysmography (PPG) is made possible monitoring minuscule periodic changes of skin color due to blood pumping through face, we conjecture normal heartbeat rhythms found in real videos will be disrupted or even entirely broken DeepFake video, making it...
With a good balance between tracking accuracy and speed, correlation filter (CF) has become one of the best object frameworks, based on which many successful trackers have been developed. Recently, spatially regularized CF (SRDCF) developed to remedy annoying boundary effects tracking, thus further boosting performance. However, SRDCF uses fixed spatial regularization map constructed from loose bounding box its performance inevitably degrades when target or background show significant...
Shadow removal is still a challenging task due to its inherent background-dependent <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup> and spatial-variant properties, leading unknown diverse shadow patterns. Even powerful deep neural networks could hardly recover traceless shadow-removed background. This paper proposes new solution for this by formulating it as an exposure fusion problem address the challenges. Intuitively, we first...
Most existing studies on next location recommendation propose to model the sequential regularity of check-in sequences, but suffer from severe data sparsity issue where most locations have fewer than five following locations. To this end, we an Attentional Recurrent Neural Network (ARNN) jointly both and transition regularities similar (neighbors). In particular, first design a meta-path based random walk over novel knowledge graph discover neighbors heterogeneous factors. A recurrent neural...
With the recent advances in voice synthesis, AI-synthesized fake voices are indistinguishable to human ears and widely applied produce realistic natural DeepFakes, exhibiting real threats our society. However, effective robust detectors for synthesized still their infancy not ready fully tackle this emerging threat. In paper, we devise a novel approach, named DeepSonar, based on monitoring neuron behaviors of speaker recognition (SR) system, i.e., deep neural network (DNN), discern voices....
Single-image deraining is rather challenging due to the unknown rain model. Existing methods often make specific assumptions of model, which can hardly cover many diverse circumstances in real world, compelling them employ complex optimization or progressive refinement. This, however, significantly affects these methods' efficiency and effectiveness for efficiency-critical applications. To fill this gap, paper, we regard single-image as a general image-enhancing problem originally propose...
The Visual Object Tracking challenge VOT2021 is the ninth annual tracker benchmarking activity organized by VOT initiative. Results of 71 trackers are presented; many state-of-the-art published at major computer vision conferences or in journals recent years. was composed four sub-challenges focusing on different tracking domains: (i) VOT-ST2021 focused short-term RGB, (ii) VOT-RT2021 "real-time" (iii) VOT-LT2021 long-term tracking, namely coping with target disappearance and reappearance...
Although achieving significant progress, existing deep generative inpainting methods still show low generalization across different scenes. As a result, the generated images usually contain artifacts or filled pixels differ greatly from ground truth, making them far real-world applications. Image-level predictive filtering is widely used restoration technique by predicting suitable kernels adaptively according to input Inspired this inherent advantage, we explore possibility of addressing...
Full face synthesis and partial manipulation by virtue of the generative adversarial networks (GANs) its variants have raised wide public concerns. In multi-media forensics area, detecting ultimately locating image forgery has become an imperative task. this work, we investigate architecture existing GAN-based methods observe that imperfection upsampling therewithin could be served as important asset for GAN-synthesized fake detection localization. Based on basic observation, proposed a...
Deep learning has recently been widely applied to many applications across different domains, e.g., image classification and audio recognition. However, the quality of Neural Networks (DNNs) still raises concerns in practical operational environment, which calls for systematic testing, especially safety-critical scenarios. Inspired by software a number structural coverage criteria are designed proposed measure test adequacy DNNs. due blackbox nature DNN, existing difficult interpret, making...
Continuous sign language recognition (CSLR) aims to recognize glosses in a video. State-of-the-art methods typically have two modules, spatial perception module and temporal aggregation module, which are jointly learned end-to-end. Existing results [9, 20, 25, 36] indicated that, as the frontal component of over-all model, used for feature extraction tends be insufficiently trained. In this paper, we first conduct empirical studies show that shallow allows more thor-ough training module....
At this moment, GAN-based image generation methods are still imperfect, whose upsampling design has limitations in leaving some certain artifact patterns the synthesized image. Such can be easily exploited (by recent methods) for difference detection of real and GAN-synthesized images. However, existing put much emphasis on patterns, which become futile if such were reduced.
The existence of motion blur can inevitably influence the performance visual object tracking. However, in contrast to rapid development trackers, quantitative effects increasing levels on trackers still remain unstudied. Meanwhile, although image-deblurring produce visually sharp videos for pleasant perception, it is also unknown whether tracking benefit from image deblurring or not. In this paper, we present a Blurred Video Tracking (BVT) benchmark address these two problems, which contains...
Motion blur caused by the moving of object or camera during exposure can be a key challenge for visual tracking, affecting tracking accuracy significantly. In this work, we explore robustness trackers against motion from new angle, i.e., adversarial attack (ABA). Our main objective is to online transfer input frames their natural motion-blurred counterparts while misleading state-of-the-art process. To end, first design synthesizing method based on generation principle blur, considering...
Spatial regularization (SR) is known as an effective tool to alleviate the boundary effect of correlation filter (CF), a successful visual object tracking scheme, from which number state-of-the-art trackers can be stemmed. Nevertheless, SR highly increases optimization complexity CF and its target-driven nature makes spatially-regularized may easily lose occluded targets or surrounded by other similar objects. In this paper, we propose selective spatial (SSR) for CF-tracking scheme. It...
As the GAN-based face image and video generation techniques, widely known as DeepFakes, have become more matured realistic, there comes a pressing urgent demand for effective DeepFakes detectors. Motivated by fact that remote visual photoplethysmography (PPG) is made possible monitoring minuscule periodic changes of skin color due to blood pumping through face, we conjecture normal heartbeat rhythms found in real videos will be disrupted or even entirely broken DeepFake video, making it...
Nowadays, digital facial content manipulation has become ubiquitous and realistic with the success of generative adversarial networks (GANs), making face recognition (FR) systems suffer from unprecedented security concerns. In this paper, we investigate introduce a new type attack to evade FR by manipulating content, called morphing (a.k.a. Amora). contrast noise that perturbs pixel intensity values adding human-imperceptible noise, our proposed works at semantic level pixels spatially in...
The state-of-the-art deep neural networks (DNNs) are vulnerable against adversarial examples with additive random-like noise perturbations. While such hardly found in the physical world, image blurring effect caused by object motion, on other hand, commonly occurs practice, making study of which greatly important especially for widely adopted real-time processing tasks (e.g., detection, tracking). In this paper, we initiate first step to comprehensively investigate potential hazards blur...
Although recent personalization methods have democratized high-resolution image synthesis by enabling swift concept acquisition with minimal examples and lightweight computation, they also present an exploitable avenue for highly accessible backdoor attacks. This paper investigates a critical unexplored aspect of text-to-image (T2I) diffusion models - their potential vulnerability to attacks via personalization. By studying the prompt processing popular (epitomized Textual Inversion...