Annotated datasets for waste electrical and electronic equipment identification and battery detection and...

Area of Science:

• Robotics and Artificial Intelligence
• Circular Economy
• Waste Management

Background:

• The circular economy demands efficient recycling of waste electrical and electronic equipment (WEEE).
• Battery fires pose significant risks in WEEE recycling facilities.
• The advanced robotics and artificial intelligence for critical raw materials recycling in the circular economy (XBAT+) project aims to address these challenges.

Purpose of the Study:

• To release initial annotated datasets for battery waste electrical and electronic equipment (WEEE).
• To enable the development of AI models for automated WEEE sorting and identification.
• To support the mitigation of fire hazards in recycling operations.

Main Methods:

• Sorting of small battery WEEE devices from general WEEE streams.
• Categorization of 421 battery WEEE devices into 15 distinct categories.
• Acquisition of 3045 static images (optical and x-ray) and semi-automatic annotation.
• Splitting datasets into training (80%) and testing (20%) sets for model evaluation.

Main Results:

• Evaluation using YOLO26 and RF-DETR demonstrated promising performance in battery-WEEE identification.
• High mAP@50 and recall achieved for supervised learning-based detection.
• The datasets show potential for effective WEEE filtering and hazard reduction.

Conclusions:

• The XBAT+ annotated datasets are valuable resources for advancing AI in WEEE recycling.
• Supervised learning models show promise for automated battery-WEEE detection.
• These datasets can significantly contribute to safer and more efficient recycling processes.