A5038283983- Generative Adversarial Networks and Image Synthesis
- Face recognition and analysis
- Advanced Image Processing Techniques
- Anomaly Detection Techniques and Applications
- Machine Fault Diagnosis Techniques
- Advanced Computing and Algorithms
- Digital Media Forensic Detection
- Human Pose and Action Recognition
- Retinal Imaging and Analysis
- Music and Audio Processing
- Neuroinflammation and Neurodegeneration Mechanisms
- Cell Image Analysis Techniques
- Single-cell and spatial transcriptomics
- Gaze Tracking and Assistive Technology
Shenzhen University
2021-2023
Shenzhen Academy of Robotics
2021
Instituto da Visão
2021
Institut de la Vision
2021
Generative Adversarial Networks (GANs) have shown compelling results in various tasks and applications recent years. However, mode collapse remains a critical problem GANs. In this paper, we propose novel training pipeline to address the issue of Different from existing methods, generalize discriminator as feature embedding maximize entropy distributions space learned by discriminator. Specifically, two regularization terms, i.e., Deep Local Linear Embedding (DLLE) Isometric Mapping...
Significant progress has been made in high-resolution and photo-realistic image generation by Generative Adversarial Networks (GANs). However, the process is still lack of control, which crucial for semantic face editing. Furthermore, it remains challenging to edit target attributes preserve identity at same time. In this paper, we propose SSFlow achieve identity-preserved manipulation StyleGAN latent space based on conditional Neural Spline Flows. To further improve performance Flows such...
In this paper, we consider the lifelong age progression and regression task, which requires to synthesize a persons appearance across wide range of ages. We propose simple yet effective learning framework achieve by exploiting prior knowledge faces captured well-trained generative adversarial networks (GANs). Specifically, first utilize pretrained GAN face images with different ages, then learn model conditional aging process in latent space. Moreover, also introduce cycle consistency loss...
Generative Adversarial Networks (GANs) have shown compelling results in various tasks and applications recent years. However, mode collapse remains a critical problem GANs. In this paper, we propose novel training pipeline to address the issue of Different from existing methods, generalize discriminator as feature embedding maximize entropy distributions space learned by discriminator. Specifically, two regularization terms, i.e., Deep Local Linear Embedding (DLLE) Isometric Mapping...
Generative Adversarial Networks (GANs) have been widely adopted in various fields. However, existing GANs generally are not able to preserve the manifold of data space, mainly due simple representation discriminator for real/generated data. To address such open challenges, this paper proposes Manifold-preserved (MaF-GANs), which generalize Wasserstein into high-dimensional form. Specifically, improve data, MaF-GANs is designed map a manifold. Furthermore, stabilize training MaF-GANs, an...
Gaze estimation often requires a large scale datasets with well annotated gaze information to train the estimator. However, such dataset costive annotation and is usually very difficult collect. Therefore, number of redirection approaches have been proposed address problem. existing methods lack ability precisely synthesize images target head pose in complex lighting scenes. As powerful technique model distribution given data, normalizing flows generate photo-realistic provide flexible...