Study on the real-time object detection approach for end-of-life battery-powered electronics in the waste of electrical and electronic equipment recycling process.

• Real-time object detection can be utilized to pre-separate EEEs. • Among 37 types of EEEs, 12 types of battery-powered EEE were sorted. • Image augmentation and transfer learning improved training and test efficiency. • YOLOv4-based backbone, precision rate showed ranging from 84.5% to 90.1%. With the growing use of electrical and electronic equipment (EEE), managing end-of-life EEE has become critical. Thus, the demand for sorting and detaching batteries from EEE in real time has increased. In this study, we investigated real-time object detection for sorting EEE, which using batteries, among numerous EEEs. To select products with batteries that have been mainly recycled, we crowd-sourced and gathered about 23,000 image datasets of the EEE with battery. Two learning techniques—data augmentation and transfer learning—were applied to resolve the limitations of the real-world data. We conducted YOLOv4-based experiments on the backbone and the resolution. Moreover, we defined this task as a binary classification problem; therefore, we recalculated the average precision (AP) scores from the network through postprocessing. We achieved battery-powered EEE detection scores of 90.1% and 84.5% at AP scores of 0.50 and 0.50–0.95, respectively. The results showed that this approach can provide practical and accurate information in the real world, hence encouraging the use of deep learning in the pre-sorting stage of the battery-powered EEE recycling industry. [ABSTRACT FROM AUTHOR]