Foreign objects detection using deep learning techniques for graphic card assembly line.

Area of Science:

• Manufacturing Engineering
• Computer Vision
• Artificial Intelligence

Background:

• Quality control in manufacturing is crucial to prevent defects and ensure product integrity.
• Foreign object detection on assembly lines, particularly for electronics like graphics cards, is vital for maintaining high production standards.

Purpose of the Study:

• To develop and evaluate deep learning models for automated foreign object detection and segmentation in graphics card assembly.
• To enhance the efficiency and accuracy of quality control processes on manufacturing lines.

Main Methods:

• Utilized convolutional neural network (CNN) models, specifically Inception Resnet v2 for classification and Attention Residual U-net++ for segmentation.
• Employed both benchmark datasets and a specific case study dataset from a graphics card assembly line for comprehensive model evaluation.

Main Results:

• The proposed CNN models demonstrated superior performance in detecting and segmenting foreign objects compared to existing methods.
• The system successfully identified and marked foreign objects, indicating its potential for real-world application.

Conclusions:

• Deep learning models, including Inception Resnet v2 and Attention Residual U-net++, offer a promising approach for automated foreign object detection in industrial assembly.
• The developed models can significantly improve the quality control of graphics card manufacturing by ensuring contaminant-free products.