A5074864253- Video Surveillance and Tracking Methods
- Face recognition and analysis
- Face and Expression Recognition
- Gaze Tracking and Assistive Technology
- Advanced Image Processing Techniques
- Machine Learning and Algorithms
- Domain Adaptation and Few-Shot Learning
- EEG and Brain-Computer Interfaces
- Hand Gesture Recognition Systems
- Advanced Steganography and Watermarking Techniques
- Image Processing Techniques and Applications
- Remote-Sensing Image Classification
- Remote Sensing and Land Use
- Emotion and Mood Recognition
- Sparse and Compressive Sensing Techniques
- Image and Video Quality Assessment
- Sentiment Analysis and Opinion Mining
- Tactile and Sensory Interactions
- Manufacturing Process and Optimization
- Color perception and design
- Chaos-based Image/Signal Encryption
- Optical Systems and Laser Technology
- Vestibular and auditory disorders
- Biometric Identification and Security
- Retinal Imaging and Analysis
Guangdong Polytechnic Normal University
2018-2024
Sun Yat-sen University
2016
Deep convolutional neural networks have been successfully applied to face detection recently. Despite making remarkable progress, most of the existing methods only localize each using a bounding box, which cannot segment from background image simultaneously. To overcome this drawback, we present and segmentation method based on improved Mask R-CNN, named G-Mask, incorporates into one framework aiming obtain more fine-grained information face. Specifically, in proposed method, ResNet-101 is...
Background: Mental health issues are increasingly prominent worldwide, posing significant threats to patients and deeply affecting their families social relationships. Traditional diagnostic methods subjective delayed, indicating the need for an objective effective early diagnosis method. Methods: To this end, paper proposes a lightweight detection method multi-mental disorders with fewer data sources, aiming improve procedures enable patient detection. First, proposed takes...
Despite the successful applications of unsupervised sparse dimensionality reduction (USDR) in pattern recognition, USDR still suffers from two challenges for hyperspectral images (HSIs), which limit its discriminative performance: first, it cannot be applied using both training samples and testing samples; second, lacks ability to integrate spectral with spatial information improving performance HSIs. In order tackle first challenge, we extend a supervised scenario, can samples, namely...
Predicting gaze point on mobile devices without calibration in unconstrained environments has great significance human computer interaction. Appearance-based estimation methods have been improved due to the recent advance convolutional neural network (CNN) models and availability of large-scale datasets. CNN limitations extracting global information features ignore important local features. In this paper, we propose a novel structure named GazeAttentionNet. To improve accuracy estimation,...
Ambient Occlusion (AO) is a technique to approximate the effect of environment lighting and add realism scene by accentuating surface details adding soft shadows, which widely used in multimedia applications. Neural Network (NNAO) pioneer introducing deep learning accurate real-time AO, but it has two limitations: 1) performance bottleneck under excessive amount samples ; 2) low contrast blurred edges leading unreal effect. To overcome these limitations, we propose Accelerated...
Existing sparse representation-based visual tracking methods detect the target positions by minimizing reconstruction error. However, due to complex background, illumination change, and occlusion problems, these are difficult locate properly. In this article, we propose a novel method based on weighted discriminative dictionaries pyramidal feature selection strategy. First, utilize color features texture of training samples obtain multiple dictionaries. Then, use position information those...