Positioning method and device of PCBA image, electronic equipment and storage medium
By combining an improved convolutional neural network with lightweight feature extraction and channel attention modules, the target detection model solves the problem of insufficient utilization of multimodal information in PCBA inspection, realizes accurate positioning and rapid detection of components, and improves detection accuracy and robustness.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- WUXI UNICOMP TECH
- Filing Date
- 2026-04-15
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies cannot effectively utilize the multimodal information of navigation maps and X-ray images in PCBA inspection, resulting in inaccurate component positioning, low inspection accuracy in complex backgrounds, and complex models with high computational load, making it difficult to conduct rapid inspection on industrial production lines.
The first and second object detection models are used to detect navigation maps and scanned images, respectively. An improved convolutional neural network is used for feature extraction and recognition. A detection model adapted to navigation maps and scanned images is constructed by combining a lightweight feature extraction module and a channel attention module.
It improves robustness and generalization ability in complex environments, ensures the reliability of component identification and classification, and meets the rapid testing needs of industrial production lines.
Smart Images
Figure CN122415478A_ABST