A5089355716- Industrial Vision Systems and Defect Detection
- Advanced Neural Network Applications
- Surface Roughness and Optical Measurements
- Smart Grid Security and Resilience
- Nanoplatforms for cancer theranostics
- Medical Imaging Techniques and Applications
- Image and Object Detection Techniques
- Nanoparticle-Based Drug Delivery
- Smart Grid and Power Systems
- Cancer, Stress, Anesthesia, and Immune Response
- Digital Radiography and Breast Imaging
- Infrastructure Maintenance and Monitoring
- AI in cancer detection
- Industrial Automation and Control Systems
Guangdong University of Technology
2021-2023
Sun Yat-sen University
2022
Tongji University
2022
Abstract Manual or conventional image processing algorithms are commonly used to detect surface problems on mobile phone screens. However, inefficiency and inflexibility disadvantages. Although the semantic segmentation method has high adaptability accuracy, it also a low defect detection efficiency due its excessive parameters. In order increase efficiency, novel efficient encoder–decoder architecture termed MB based MBConv blocks, that reduces number of parameters in methods i presented....
Objective. In digital breast tomosynthesis (DBT), architectural distortion (AD) is a lesion that difficult to detect. Compared with typical ADs, which have radial patterns, identifying ADs more difficult. Most existing computer-aided detection (CADe) models focus on the of ADs. This study focuses atypical and develops deep learning-based CADe model an adaptive receptive field in DBT.Approach. Our proposed uses Gabor filter convergence measure depict distribution fibroglandular tissues DBT...
Surface defect detection is essential for ensuring product quality. Intelligent widely studied and applied in many industrial fields. However, glass a daunting task because the optical properties of present unique challenges, e.g., intra-class difference, low contrast, ambiguous edges. In this paper, we propose an efficient edge enhancement network (EEE-Net) to address above challenges. EEE-Net employs Efficient Transformer Blocks compose pyramid network; each block combines sequence...
Defect detection in mobile screens is a pivotal aspect of the manufacturing industry. In this paper, we present novel approach for screen defect based on U-Net network named DSU-Net. We observe that conventional encoder composed only few convolutional layers, which impedes network's capability to comprehensively capture crucial target features. Additionally, direct connection between bottom and decoder U-shaped structure increases computational complexity, making it unsuitable real-time...
In this paper, a new semantic segmentation algorithm, EU-Net (Efficient U-Net), is proposed to realize surface defect detection of mobile phone screens. Compared with U-Net, the encoder and decoder are modified EfficientNet-B0 MBconv Block enhance efficiency accuracy. Due loss feature information in cropping operation, it removed our improve addition, conventional image processing techniques used dataset. The experiments conducted on dataset collected from production site screens verify...