A5101804035- Generative Adversarial Networks and Image Synthesis
- Digital Media Forensic Detection
- Face recognition and analysis
- Adversarial Robustness in Machine Learning
- Anomaly Detection Techniques and Applications
- Advanced Steganography and Watermarking Techniques
- Advanced Image Processing Techniques
- Advanced Malware Detection Techniques
- Biometric Identification and Security
- Electromagnetic Scattering and Analysis
- Marine and coastal ecosystems
- Aquatic Ecosystems and Phytoplankton Dynamics
- Soil and Water Nutrient Dynamics
- Aesthetic Perception and Analysis
- Cancer therapeutics and mechanisms
- Fault Detection and Control Systems
- Chaos-based Image/Signal Encryption
- Scientific Research and Discoveries
- Software Testing and Debugging Techniques
- Phenothiazines and Benzothiazines Synthesis and Activities
- Matrix Theory and Algorithms
- Chromatography in Natural Products
- Human Pose and Action Recognition
- Image and Video Quality Assessment
- Traditional Chinese Medicine Analysis
National University of Defense Technology
2024-2025
University of Science and Technology of China
2020-2024
Beijing Normal University
2024
Shandong University of Traditional Chinese Medicine
2023
Sichuan University
2023
Chinese Academy of Sciences
2022
Tsinghua University
1997-2000
Face swapping has drawn a lot of attention for its compelling performance. However, current deepfake methods suffer the effects obscure workflow and poor To solve these problems, we present DeepFaceLab, dominant framework practical face-swapping. It provides necessary tools as well an easy-to-use way to conduct high-quality also offers flexible loose coupling structure people who need strengthen their pipeline with other features without writing complicated boilerplate code. We detail...
Deep neural networks have made tremendous progress in 3D point-cloud recognition. Recent works shown that these recognition are also vulnerable to adversarial samples produced from various attack methods, including optimization-based Carlini-Wagner attack, gradient-based iterative fast gradient method, and skeleton-detach based point-dropping. However, after a careful analysis, methods either extremely slow because of the optimization/iterative scheme, or not flexible support targeted...
Deepfake defense not only requires the research of detection but also efforts generation methods. However, current deepfake methods suffer effects obscure workflow and poor performance. To solve this problem, we present DeepFaceLab, dominant framework for face-swapping. It provides necessary tools as well an easy-to-use way to conduct high-quality offers a flexible loose coupling structure people who need strengthen their pipeline with other features without writing complicated boilerplate...
Recently, Li et al. proposed an effective non-additive distortion model for JPEG steganography by preserving Block Boundary Continuity (BBC) in the spatial domain. However, method based on BBC only explored how modifications of DCT coefficient pairs at same mode adjacent blocks will impact BBC. In this paper, we propose a to enhance BBC, called BBC++, considering mutual from all coefficients blocks. To do that, design updating strategies both covers and costs multi-round embedding...
Recently almost all the mainstream deepfake detection methods use Convolutional Neural Networks (CNN) as their backbone. However, due to overreliance on local texture information which is usually determined by forgery of training data, these CNN-based cannot generalize well unseen data. To get out predicament prior methods, in this paper, we propose a novel transformer-based framework model both global and analyze anomalies face images. In particular, design attention leading module,...
Essential hypertension (EH) is a leading cause of cardiovascular morbidity and mortality. Fructus Tribuli (FT), as traditional medicine, has been frequently used for thousands years. The crude (CFT), decoction pieces being processed to remove impurities, have listed an important medicine the treatment in elderly. According theory Chinese CFT can enhance EH after stir-fried into (SFT). At present, whether SFT its potential pharmacodynamic substances mechanism are unknown. In this study,...
Invertible grayscale is a special kind of from which the original color can be recovered. Given an input image, this seminal work tries to hide information into its counterpart while making it hard recognize any anomalies. This powerful functionality enabled by training hiding sub-network and restoring in end-to-end way. Despite expressive results, two key limitations exist: 1) The restored image often suffers some noticeable visual artifacts smooth regions. 2) It very sensitive JPEG...
As an effective method for intellectual property (IP) protection, model watermarking technology has been applied on a wide variety of deep neural networks (DNN), including speech classification models. However, how to design black-box scheme automatic recognition (ASR) models is still unsolved problem, which significant demand protecting remote ASR Application Programming Interface (API) deployed in cloud servers. Due conditional independence assumption and label-detection-based evasion...
3D-aware GANs have shown their impressive power on 3D controlling for synthesized portraits. While the plausible facial reality is achieved, inherent properties of generated results actually not been well analyzed. One reasons that wildly-used metrics, such as Inception Score (IS) or Fréchet Distance (FID), focus more perceptual features rather than explicit clues. In this article, we propose two novel which measure face consistency and diversity a level, to compensate IS FID GAN evaluation....
This paper presents a novel Physically-guided Disentangled Implicit Rendering (PhyDIR) framework for highfidelity 3D face modeling. The motivation comes from two observations: Widely-used graphics renderers yield excessive approximations against photo-realistic imaging, while neural rendering methods produce superior appearances but are highly entangled to perceive 3D-aware operations. Hence, we learn disentangle the implicit via explicit physical guidance, guaranteeing properties of: (1)...
Face swapping is an intriguing and intricate task in the field of computer vision. Currently, most mainstream face methods employ recognition models to extract identity features inject them into generation process. Nonetheless, such often struggle effectively transfer information, which leads generated results failing achieve a high similarity source face. Furthermore, if we can accurately disentangle controllable swapping, thereby providing more choices users. In pursuit this goal, propose...
Deepfake aims to swap a face of an image with someone else’s likeness in reasonable manner. Existing methods usually perform deepfake frame by frame, thus ignoring video consistency and producing incoherent results. To address such problem, we propose novel framework Neural Identity Carrier (NICe), which learns identity transformation from arbitrary face-swapping proxy via U-Net. By modeling the incoherence between frames as noise, NICe naturally suppresses its disturbance preserves primary...
DprE1 has been characterized as one of the most attractive targets for anti-tuberculosis drug discovery. Benzothiazinones (BTZs), inhibitors, have attracted attention more and medicinal chemists. However, metabolic stability genotoxic issues BTZs may restrict their further development. We obtained a new series methylated-benzothiazinones (cBTZs) through introduction methyl group into BTZ core. Among which, compound 12f exhibits an excellent anti-tubercular activity MIC is lower than PBTZ169...
Face swapping is an intriguing and intricate task in the field of computer vision. Currently, most mainstream face methods employ recognition models to extract identity features inject them into generation process. Nonetheless, such often struggle effectively transfer information, result generated results failing achieve a high similarity with source face. Furthermore, if we can accurately disentangle controllable swapping, thereby providing more choices users. In pursuit this goal, propose...
This paper presents the summary report on DeepFake Game Competition Visual Realism Assessment (DFGC-VRA). Deep-learning based face-swap videos, also known as deepfakes, are becoming more and realistic deceiving. The malicious usage of these videos has caused wide concerns. There is a ongoing deepfake game between its creators detectors, with human in loop. research community been focusing automatic detection fake but assessment their visual realism, perceived by eyes, still an unexplored...
A parallel algorithm for Householder Transformation is given in this paper. Based on the parallelizing of Transformation, we propose algorithms solving ill-conditioned matrix inversion, linear system and least squares. The proposed methods have acceptable execution time number employed processors even handling problems.
Deep neural networks have made tremendous progress in 3D point-cloud recognition. Recent works shown that these recognition are also vulnerable to adversarial samples produced from various attack methods, including optimization-based Carlini-Wagner attack, gradient-based iterative fast gradient method, and skeleton-detach based point-dropping. However, after a careful analysis, methods either extremely slow because of the optimization/iterative scheme, or not flexible support targeted...