A5103147572- Advanced Image and Video Retrieval Techniques
- Handwritten Text Recognition Techniques
- Vehicle License Plate Recognition
- Face recognition and analysis
- Generative Adversarial Networks and Image Synthesis
- Advanced Vision and Imaging
- Image Processing and 3D Reconstruction
- Anomaly Detection Techniques and Applications
- Video Analysis and Summarization
- Image Retrieval and Classification Techniques
- Bacillus and Francisella bacterial research
- Image Enhancement Techniques
- Video Surveillance and Tracking Methods
- Digital Media Forensic Detection
- Multimodal Machine Learning Applications
- Adversarial Robustness in Machine Learning
- Fire Detection and Safety Systems
Sun Yat-sen University
2023-2025
Shenzhen University
2025
University of Chinese Academy of Sciences
2017-2019
Institute of Information Engineering
2019
Chinese Academy of Sciences
2019
Scene text plays a significant role in image and video understanding, which has made great progress recent years. Most existing models on detection the wild have assumption that all texts are surrounded by rotated rectangle or quadrangle. While there also exist lots of curved wild, would not be bounded regular bounding box. In this paper, we develop novel architecture to localize regions, can deal with curved-shape scene texts. Specifically, first design text-related feature enhancement...
Privacy information existing in the scene text will be leaked with spread of images cyberspace. Vanishing from image is a simple yet effective method to prevent privacy disclosure machine and human. Previous visual vanishing methods have achieved promising results but performance still fell short expectations for complicated-shape texts various scales. In this paper, we propose novel hierarchical context-aware interaction reconstruction make vanish natural image. To avoid interference...
Scene text detection is currently a popular research topic in the computer vision community. However, it challenging task due to variations of texts and clutter backgrounds. In this paper, we propose novel framework for scene localization. Based on region proposal network, Strip-based Text Detection Network (STDN) developed with vertical anchor mechanism predict text/non-text strip-shaped proposals. Meanwhile, incorporate recurrent neural network layers proposed refine predicted results....
Many studies demonstrate that supervised learning techniques are vulnerable to adversarial examples. However, threats in unsupervised have not drawn sufficient scholarly attention. In this article, we formally address the unexplored attacks equally important clustering field and propose concept of set attack for clustering. To illustrate basic idea, design a novel space-mapping algorithm confuse subspace clustering, one mainstream branches It maps sample into wrong class by moving it towards...
Scene text editing aims to replace the source with target while preserving original background. Its practical applications span various domains, such as data generation and privacy protection, highlighting its increasing importance in recent years. In this study, we propose a novel Text Editing network Explicitly-decoupled transfer Minimized background reconstruction, called STEEM. Unlike existing methods that usually fuse style, content, background, our approach focuses on decoupling style...