A5100685735- Astrophysics and Cosmic Phenomena
- Gamma-ray bursts and supernovae
- Video Surveillance and Tracking Methods
- Advanced Vision and Imaging
- Advanced Neural Network Applications
- Advanced Image and Video Retrieval Techniques
- Image and Signal Denoising Methods
- Remote Sensing and Land Use
- Video Analysis and Summarization
- Pulsars and Gravitational Waves Research
- Remote Sensing and LiDAR Applications
- Particle Detector Development and Performance
- Image Enhancement Techniques
- Autonomous Vehicle Technology and Safety
- Generative Adversarial Networks and Image Synthesis
- Computer Graphics and Visualization Techniques
- Image Retrieval and Classification Techniques
- Advanced Clustering Algorithms Research
- Radio Astronomy Observations and Technology
- Multimodal Machine Learning Applications
- 3D Shape Modeling and Analysis
- Photopolymerization techniques and applications
- Dark Matter and Cosmic Phenomena
- Chaos-based Image/Signal Encryption
- Human Mobility and Location-Based Analysis
Shanghai Jiao Tong University
2024-2025
Shandong University of Science and Technology
2019-2023
The University of Sydney
2023
Academy of Opto-Electronics
2022
BOE Technology Group (China)
2022
Inner Mongolia University
2021
National University of Mongolia
2021
Shanghai University of Electric Power
2021
Zhejiang University
2020
Beihang University
2009
Multi-Target Multi-Camera tracking is a fundamental task for intelligent traffic systems. The track 1 of AI City Challenge 2022 aims at the city-scale multi-camera vehicle task. In this paper we propose an accurate system composed 4 parts, including: (1) State-of-the-art detection and re-identification models feature extraction. (2) Single camera tracking, where introduce augmented tracks prediction multi-level association method on top tracking-by-detection paradigm.(3) Zone-based...
Hyperspectral LiDAR (HSL) has been utilised as an efficacious technique in objects classification and recognition based on its synchronously obtaining spectral spatial information. However, the information obtained by most of developed HSL was visible near-infrared range (VNIR, 400–1000 nm). Whereas a longer wavelength showed more useful for detection, such detecting vegetation water content. This paper proposed tested eight-channel prototype covering to even short-wavelength infrared...
The perspective distortion of oblique images has a substantial impact on dense matching, i.e., it reduces the matching precision. In this article, strategy in which object plane direction is considered proposed. According to many regular planes urban areas, epipolar rectification with minimum distortions relative selected reference can be generated. results various are weighted and fused into single depth map, better result. experimental demonstrate that influence performance....
In order to solve the problem of information loss, some image steganography methods utilize generative adversarial networks (GANs), while existing can not capture both texture and semantic features. this paper, a more accurate method is proposed, where multi-level feature fusion procedure based on GAN designed. Firstly, convolution pooling operations are added network for extraction. Then, short links used fuse information. Finally, stego generated by confrontation learning between...
Common depth image fusion methods use each original as a reference plane and fuse the images using mutual projection. These can eliminate inconsistency between images, but they cannot alleviate point cloud redundancy computational complexity. This article proposes virtual view method for fusion, defines limited number of views by means clustering, reduces redundant calculations, covers all scenes much possible. The is merged ray ray, reliable value obtained via F-test. Compared with modified...
In the substation monitoring environment, helmet wearing of workers is an important approach to ensure safety. Because size in entire surveillance images often small and characteristic information unclear, existing target detection algorithms present problem missed omission. A safety dynamic model based on critical area attention mechanism first detected human image, then, locked head through network. Finally, feature map areas was up-sampled many times increase proportion image highlight...
The paper proposes a real-time subsurface scattering rendering method for dynamic objects. Subsurface owned by translucent objects is complex optical phenomenon. Since the complexity of property description materials and their interaction with light, pre-computing methods are usually used to render objects, but these approaches either limit be non-deformable or homogeneous. In this paper, new material describing decomposed light computing model set forth, on basis which image-space...
Detecting signals that are long, weak, and narrowband is a well known important problem in acoustic signal processing. In this paper an ad hoc scheme developed: its stages include the DFT, multiresolution decomposition frequency domain, GLRT. The computational load light, performance remarkably good. This so not just original situation, but also, due to inherent adaptivity data, detection of relatively broadband nature. Generalizations given CFAR operation both prewhitened unwhitened cases,...
Pedestrian detection is an important branch of object detection. It plays a vital role in many fields such as intelligent monitoring systems. The premise the pedestrian recognition algorithm application industrial scene accurate Our paper proposes model to solve real-time with high accuracy base on YOLO v3. We provide method select size and number anchor boxes for predicting bounding accurately. Then we use modified shuffle unit lightweight backbone v3, which reduces 67.3% FLOPs 65.1%...