.png){kind=icon}
.png){kind=icon}
Return
Efficient Micro-LED defect detection based on microscopic vision and deep learning
.png){kind=icon}
PRE
AI
DOI：10.1016/j.optlaseng.2024.108116.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Defect detection is highly important in the production of Micro-LEDs. Micro-LEDs are characterized by their small size and high density, which leads to a variety of complex defects. Existing detection methods often fail to meet the accuracy and speed requirements of defect detection. To address these issues, we constructed an automatic microscopic vision-based detection system (MVDS) for capturing images of Micro-LED chips and proposed a deep learning algorithm called LBG-YOLO, which recognizes and locates multiple types of defects while meeting real-time detection requirements. First, a lightweight multiscale convolutional module (LMCM) is proposed to improve feature extraction and feature fusion capabilities. Second, a bidirectional feature pyramid network (BiFPN) is used in the neck network to fuse features more effectively while reducing the size of the network. Then, a global dependency coordinate attention (GDCA) mechanism is designed to make full use of the location information and locate the object more accurately. The experimental results on the Micro-LED dataset and public VOC2017 dataset show that LBG-YOLO achieves competitive results with those of other state-of-theart methods, with optimal accuracy (90.8 % mAP).
Keywords:
Micro-LED
Defect detection
Microscopic vision-based
Convolutional neural network
Lightweight convolution
Self-attention
Journal
IF:
3.7
Papers:
7.2K
Citations:
1.7W