A5025145926- Industrial Vision Systems and Defect Detection
- Advanced Neural Network Applications
- Digital Radiography and Breast Imaging
- Image and Object Detection Techniques
- Dental Radiography and Imaging
- Advanced Image and Video Retrieval Techniques
- Remote-Sensing Image Classification
- Advanced Image Fusion Techniques
Systems Engineering Society of China
2023
Shanghai Institute of Technology
2022
Arbitrary-Oriented object detection in remote sensing images is a hot topic recent years. Currently, most arbitrary-oriented detectors adopt the oriented bounding box (OBB) to represent targets imagery. However, OBB representation suffers from suboptimal regression problems caused by ambiguity of angle definition. In this paper, we propose novel framework Learning Segmentation-aware Mask for Detection (LSM-Det) LSM-Det predicts mask object, and then converts prediction into minimum external...
PCB is an important component of electronic products, which critical in the quality production. But It has those problems including low efficiency current surface defect detection methods and high equipment cost caused by a large deep learning model. A method involves combining simplified YOLOv5s network with model channel pruning through follow three steps. First all it's based on characteristics relatively small defects manufacturing, removes convolution feature map scale Secondly,...
During the automatic reconstruction of panoramic images, effect dental arch curve fitting will affect integrity content image. Metal implants in patient's mouth usually lead to a decrease contrast image, which affects doctor's diagnosis. In this paper, an oral image method was proposed. By calculating key areas and extraction fusion algorithms, could be automatically detected adjusted on small number intensity distribution teeth, bone tissue metal reduce impact other tissues, generate...
The manufacturing process of industrial aluminum profiles is complex, resulting in many types defects and difficult detection. Therefore, we propose an improved YOLO algorithm that can effectively improve the efficiency defect First all, analyzed characteristics designed data augmentation strategies to training efficiency. Secondly, feature fusion network was improved, attention mechanism added enhance deep shallow information, detection scale algorithm's adaptability multi-scale targets,...