A computer vision-based system for real-time component identification from waste printed circuit boards.

Area of Science:

• Environmental Science
• Computer Science
• Materials Science

Background:

• Electronic waste (e-waste) poses significant economic and environmental challenges due to increasing consumer demand for electronics.
• Printed Circuit Boards (PCBs) are a major component of e-waste, containing valuable yet hazardous materials.
• Manual segregation of Electronic Components (ECs) from PCBs in recycling processes is inefficient, impacts human health, and leads to material loss.

Purpose of the Study:

• To develop an automated system for classifying and segregating Electronic Components (ECs) from waste Printed Circuit Boards (PCBs).
• To address the limitations of manual sorting in e-waste recycling, particularly in developing nations.
• To enhance the efficiency and safety of e-waste recycling processes through technological intervention.

Main Methods:

• Implementation of a computer vision and deep learning approach, specifically the YOLOv3 algorithm, for EC identification.
• Utilizing a combination of a publicly available dataset and a custom PCB dataset to train the model.
• Testing the system under challenging real-world recycling conditions, including varied lighting, shadows, orientations, viewpoints, and resolutions.

Main Results:

• The YOLOv3 detection model demonstrated satisfactory classification accuracy for ECs.
• The system exhibited real-time identification capabilities, crucial for automated sorting.
• The developed system shows potential for effective segregation of ECs from waste PCBs.

Conclusions:

• The proposed computer vision and deep learning system offers a robust solution for automated EC identification and segregation.
• This automation can significantly improve the efficiency of e-waste recycling, leading to better recovery of valuable materials.
• Adopting such automated solutions is vital for mitigating the environmental impact of e-waste and protecting worker health.