Daily Arxiv

This page organizes papers related to artificial intelligence published around the world.
This page is summarized using Google Gemini and is operated on a non-profit basis.
The copyright of the paper belongs to the author and the relevant institution. When sharing, simply cite the source.

Image Segmentation and Classification of E-waste for Training Robots for Waste Segregation

Created by
  • Haebom

Author

Prakriti Tripathi

Outline

This paper explores the use of machine learning models to solve the e-waste classification problem presented by an industry partner. We built our own dataset by disassembling and photographing common e-waste items, such as mice and chargers, and trained the state-of-the-art YOLOv11 and Mask-RCNN models. As a result, the YOLOv11 model achieved 70 mAP in real-time, while the Mask-RCNN model achieved 41 mAP. The trained models can be integrated with a picking-and-place robot and utilized for e-waste classification tasks.

Takeaways, Limitations

•
Takeaways:
◦
Development of a real-time object detection model for electronic waste classification and its applicability are presented.
◦
The excellent performance of the YOLOv11 model demonstrates its applicability to the field of electronic waste classification.
◦
Integration with picking and placing robots presents the potential to contribute to improving the efficiency of e-waste recycling.
•
Limitations:
◦
There is a need to verify the generalizability of self-built datasets.
◦
Lack of model performance evaluation for various types of e-waste and complex background environments.
◦
The performance of the Mask-RCNN model is relatively low compared to the YOLOv11 model.
◦
Lack of details on integration with actual robotic systems and field applications.
👍
Made with Slashpage