A5091472897- Generative Adversarial Networks and Image Synthesis
- Advanced Image Processing Techniques
- Digital Media Forensic Detection
- Robotics and Sensor-Based Localization
- Face recognition and analysis
- Advanced Image and Video Retrieval Techniques
- Advanced Neural Network Applications
Meizu (China)
2023
Institute of Computing Technology
2019-2020
Chinese Academy of Sciences
2019-2020
University of Chinese Academy of Sciences
2019
Combined variations containing low-resolution and occlusion often present in face images the wild, e.g., under scenario of video surveillance. While most existing image recovery approaches can handle only one type variation per model, this work, we propose a deep generative adversarial network (FCSR-GAN) for performing joint completion super-resolution via multi-task learning. The generator FCSR-GAN aims to recover high-resolution without given an input with occlusion. discriminator uses set...
Occlusions are often present in face images the wild, e.g., under video surveillance and forensic scenarios. Existing de-occlusion methods limited as they require knowledge of an occlusion mask. To overcome this limitation, we propose paper a new generative adversarial network (named OA-GAN) for natural without mask, enabled by learning semi-supervised fashion using (i) paired with known masks artificial occlusions (ii) masks. The generator our approach first predicts which is used filtering...
Combined variations such as low-resolution and occlusion often present in face images the wild, e.g., under scenario of video surveillance. While most existing enhancement approaches only handle one type variation per model, this paper, we propose a deep generative adversarial network (FCSR-GAN) for joint completion super-resolution via model. The generator FCSR-GAN aims to recover high-resolution image without given an input with partial occlusions. discriminator consists two losses,...
Cross-view geo-localization is to retrieve images of the same geographic target from different platforms. Since drones have received increasing attention in recent years because their ability capture high-quality multimedia data sky, we focus on image retrieval drone platform satellite this paper. We propose an attention-guided feature partition network (AFPN) which leverages learnable spatial maps divide global high-level map into class-aware foreground and class-agnostic background...
Combined variations containing low-resolution and occlusion often present in face images the wild, e.g., under scenario of video surveillance. While most existing image recovery approaches can handle only one type variation per model, this work, we propose a deep generative adversarial network (FCSR-GAN) for performing joint completion super-resolution via multi-task learning. The generator FCSR-GAN aims to recover high-resolution without given an input with occlusion. discriminator uses set...