A5061406943- Video Surveillance and Tracking Methods
- Gait Recognition and Analysis
- Human Pose and Action Recognition
- Advanced Image and Video Retrieval Techniques
- Advanced Neural Network Applications
- Face recognition and analysis
- Hand Gesture Recognition Systems
- Diabetic Foot Ulcer Assessment and Management
- Vehicle License Plate Recognition
- Fire Detection and Safety Systems
- IoT-based Smart Home Systems
Beijing University of Posts and Telecommunications
2016-2019
Gait recognition is an attractive human technology. However, existing gait methods mainly focus on the regular cycles, which ignore irregular situation. In real-world surveillance, almost irregular, contains arbitrary dynamic characteristics (e.g., duration, speed, and phase) varied viewpoints. this paper, we propose attentive spatial-temporal summary networks to learn salient view-independence features for recognition. First of all, design gate mechanism with extract discriminative...
Person re-identification across multiple camera views is a rather challenging task due to various view points, illuminations, backgrounds and poses. How extract discriminative features the most critical way overcome these challenges. In this paper, we design null space based deep learning approach for person re-identification. Firstly, Siamese Convolutional Neural Network (SCNN) designed automatically learn effective semantic in different views. Furthermore, obtain better recognition...
Cross-view gait recognition is a challenging problem when view-interval and pose variation are relatively large. In this paper, we propose Cycle-consistent Attentive Generative Adversarial Networks (CA-GAN) to map different views' images view-consistent photorealistic for cross-view recognition. CA-GAN, the generative network composed of two branches, which simultaneously perceives human's global contexts local body parts information respectively. Moreover, design novel Network (AAN)...
Vehicle search aims to find a specific vehicle appeared in the physical world through surveillance networks. Existing systems usually exploit attributes, like colors and types, or license plate numbers as keywords vehicles database. However, attribute based cannot accurately target due minor inter-class difference between similar extremely varied environmental factors. Moreover, plates be correctly recognized real-world scenes viewpoints motion blur. In this paper, we design progressive...
Person re-identification has attracted increasing research interest due to its great potential ability find the target person in large-scale surveillance videos. Most existing methods for only achieve limited performance practical applications, as they mainly focus on generic appearance of while neglecting some unique identities (e.g., human gait). In this paper, we propose a progressive approach simultaneously improve timeliness and accuracy identifying person. Our is treated coarse-to-fine...
This paper is focused on the task of searching for a specific vehicle that appeared in surveillance networks. Existing methods usually assume images are well cropped from videos, then use visual attributes, like colors and types, or license plate numbers to match target image set. However, complete search system should consider problems detection, representation, indexing, storage, matching, so on. Besides, attribute-based cannot accurately find same due intra-instance changes different...