Gold detection from printed circuit boards for eco-friendly e-waste recycling

Area of Science:

• Materials Science
• Computer Science
• Environmental Science

Background:

• Electronic waste (e-waste) is the world's fastest-growing waste stream, containing valuable resources like gold.
• Current gold recovery methods from e-waste Printed Circuit Boards (PCBs) face challenges in accurately identifying gold regions.
• A lack of standardized datasets for PCBs hinders the development of efficient gold recovery processes.

Purpose of the Study:

• To develop a lightweight, Region of Interest (RoI)-guided deep learning framework for precise gold detection on PCBs.
• To address the limitations of existing methods in identifying gold-bearing areas in e-waste.
• To enhance the efficiency and sustainability of gold recovery from waste PCBs.

Main Methods:

• A novel RoI-guided deep learning framework utilizing the YOLOv9 model was developed.
• A custom dataset of PCB images was created, pre-processed, and annotated for gold content.
• Region of Interest (RoI) generation involved fusing local entropy with Hue, Saturation, Value (HSV) thresholding to guide detection.

Main Results:

• The proposed RoI-guided YOLOv9 model demonstrated robust and effective detection of gold regions on PCBs.
• The model achieved higher detection accuracy and superior localization of gold-bearing areas compared to baseline methods and other YOLO variants.
• Computational efficiency was maintained, indicating practical applicability for e-waste processing.

Conclusions:

• The developed deep learning framework offers a reliable and precise solution for detecting gold components in PCB images.
• This advancement can significantly improve the sustainability and efficiency of gold recovery from electronic waste.
• The study highlights the potential of AI-driven approaches in addressing critical challenges in e-waste management and resource recovery.