What is it about?
This paper presents an image-based classification framework integrated into the Waste Electrical and Electronic Equipment (WEEE) recycling pipeline to automate the sorting of electric motors. Sourced from an industrial recycling facility, a heterogeneous dataset of 229 electric motor samples was used to train a two-stage computer vision model. The framework first employs a Faster R-CNN to localize the Region of Interest (RoI) of the motor, and then utilizes a ResNeXt-50 convolutional neural network optimized via Discriminative Transfer Learning (DTL). The model achieves a dual objective: estimating the probability that a motor contains permanent magnets (PMs) for rare earth recovery, and classifying the motor into one of four commercially widespread construction typologies.
Featured Image
Photo by Troy Winther on Unsplash
Why is it important?
The energy transition relies heavily on Rare Earth Elements (REEs) like Neodymium, Praseodymium, and Dysprosium, which are critical raw materials for permanent magnet (PM) electric motors due to their high power density. While large industrial motors are sometimes manually dismantled, PM-containing motors from household electronics frequently escape detailed characterization. Consequently, they risk being mixed with generic metal scrap and subjected to mass shredding, which permanently dilutes and destroys these valuable materials. This study addresses a key industrial bottleneck by offering an automated, deep-learning-based classification tool. By sorting PM-containing motors prior to shredding, it lays the foundation for "Functional Recycling" at an industrial scale, preserving rare earth quality and helping facilities meet European Critical Raw Materials Act (CRMA) recovery targets.
Perspectives
Modern household appliances, like vacuum cleaners and washing machines, use high-efficiency electric motors powered by strong magnets made of rare earth elements. These materials are incredibly precious and environmentally costly to mine. Unfortunately, when electronic waste is recycled today, these small motors are typically tossed into giant shredders along with other metal scrap, which permanently destroys and wastes the rare earth metals inside. To solve this issue, researchers created an automated system that uses smart cameras and artificial intelligence (deep learning) to look at scrap motors and immediately recognize which ones contain these valuable magnets. Successfully identifying 81% of magnet-containing motors, this technology allows recycling plants to easily separate them before shredding, ensuring these critical green materials can be saved, recycled, and reused.
Maurizio Guadagno
Universita degli Studi di Firenze
Read the Original
This page is a summary of: Toward Automated Detection of Permanent Magnet Motors in WEEE Recycling Using Discriminative Transfer Learning, Machines, March 2026, MDPI AG,
DOI: 10.3390/machines14030331.
You can read the full text:
Contributors
The following have contributed to this page
