A5074190149- Advanced Neural Network Applications
- Infrared Target Detection Methodologies
- Video Surveillance and Tracking Methods
- Advanced Image and Video Retrieval Techniques
- Image Processing and 3D Reconstruction
- Robotics and Sensor-Based Localization
- Generative Adversarial Networks and Image Synthesis
- Polyomavirus and related diseases
- Visual Attention and Saliency Detection
- Cancer-related molecular mechanisms research
- Cultural Heritage Materials Analysis
- Medical Imaging and Analysis
- RNA Research and Splicing
- Dental Radiography and Imaging
- Autonomous Vehicle Technology and Safety
- RNA modifications and cancer
- Optical Systems and Laser Technology
- Medical Image Segmentation Techniques
- Nanoparticle-Based Drug Delivery
- Image Processing Techniques and Applications
- Vehicle License Plate Recognition
- 3D Surveying and Cultural Heritage
- Image Enhancement Techniques
- Digital Media Forensic Detection
- Nanoplatforms for cancer theranostics
Zhejiang Cancer Hospital
2025
China West Normal University
2023-2024
Xihua University
2023
Sichuan University
2020-2021
Chinese Academy of Medical Sciences & Peking Union Medical College
2021
Object detection is one of the core technologies for autonomous driving. Current road object mainly relies on visible light, which prone to missed detections and false alarms in rainy, night-time, foggy scenes. Multispectral based fusion RGB infrared images can effectively address challenges complex changing scenes, improving performance current algorithms scenarios. However, previous multispectral suffer from issues such as poor dual-mode information, multi-scale objects, inadequate...
Vehicle detection and tracking technology plays a crucial role in Intelligent Transportation Systems. However, due to factors such as complex scenarios, diverse scales, occlusions, issues like false detections, missed identity switches frequently occur. To address these problems, this paper proposes multi-object vehicle algorithm based on CDS-YOLOv8 improved ByteTrack. For detection, the Context-Guided (CG) module is introduced during downsampling process enhance feature extraction...
Metastasis, a powerful prognostic indicator of oral squamous cell carcinoma (OSCC), is chiefly responsible for poor cancer outcomes. Despite an increasing number studies examining the mechanisms underlying outcomes, development potent strategies hindered by insufficient characterization crucial regulators. Long noncoding RNAs (lncRNAs) have recently been gaining interest as significant modulators OSCC metastasis; however, detailed lncRNA-mediated metastasis remain relatively uncharacterized....
To address the issues of blurred edges and contours, insufficient extraction low-frequency information, unclear texture details in ancient murals, which lead to decreased ornamental value limited research significance this paper proposes a novel mural super-resolution reconstruction method, based on an attention mechanism multi-level residual network, termed AM-ESRGAN. This network builds module for Multi-Scale Dense Feature Fusion (MDFF) adaptively fuse features at different levels more...
The accurate and rapid detection of traffic signs is crucial for intelligent transportation systems. Aiming at the problems that have including more small targets in road scenes as well misdetection, omission, low recognition accuracy under influence fog, we propose a model detecting foggy scenes—YOLO-TSF. Firstly, design CCAM attention module combine it with idea local–global residual learning thus proposing LGFFM to enhance capabilities weather. Secondly, MASFFHead by introducing ASFF...
Due to the challenges of small detection targets, dense target distribution, and complex backgrounds in aerial images, existing object algorithms perform poorly image tasks. To address these issues, this paper proposes an improved algorithm called YOLOv5s-DSD based on YOLOv5s. Specifically, SPDA-C3 structure is proposed used reduce information loss while focusing useful features, effectively tackling targets backgrounds. The novel decoupled head structure, Res-DHead, introduced, along with...
In response to the challenges of small targets and complex backgrounds in remote sensing image detection, this paper proposes an improved algorithm based on Yolov5s, named Yolov5s-RSD. The replaces original downsampling method with SPD-Conv reduce information loss during downsampling, adds a detection head for fully utilize their features, introduces Biformer attention address background issues. Testing VisDrone datasets shows that achieves 10.0% increase map@0.5 7.0% map@0.5:0.95 compared algorithm.
Aiming to address the issues of small target size, complex background, and low recognition accuracy traffic signs in road scenes, an improved YOLOv5s algorithm is proposed. The introduces a Coordinate Attention feature fusion network, enhancing model's ability integrate spatial coordinate information preserve wider range positional information. Efficient Decoupled Heads are used replace Detection YOLOv5s, with additional layer for object detection, improving generalization ability,...
In this paper, we introduce an enhanced model named YOLOv5s-SMF, which is a refinement of YOLOv5s designed to address the challenges small object detection in remote sensing images, including target size, limited feature information, and complex background. It introduces SPD-Conv as re-placement for original downsampling method, reducing information loss during downsampling. Moreover, tiny layer added fully utilize features objects. Adopting multi view fusion module enhance model's...
Objective To discuss a new coding method for individual identification based on oral panoramic tomography, analyze the diversity of different modules in nonhomologous images and consistency matching rate indexes homologous images, evaluate application value identification. Methods The tomography 1 000 patients with permanent teeth were collected retrospectively. Each patient had two taken at times （called Early database Late according to chronological order）. image was coded designed method....