A modified U-Net CNN model for enhanced battery component segmentation in X-ray phone images

Abstract

This article highlights the expanding issue of e-waste caused by the accessibility and widespread utilisation of electronics. Because precious metals in e-waste have high value, it is vital to recycle them while minimising their loss. Batteries in e-waste are identified and located using image processing techniques, such as semantic segmentation, which categorizes each pixel in an image. The article describes a modified U-Net Convolutional Neural Network approach with pre-processing procedures to assure clean raw photos for image segmentation of the battery component. Key matrices were used to analyse the output of three distinct CNN models with loss functions. The study comes to the conclusion that the improved model for battery segmentation of X-ray images is the modified U-Net with dice coefficient. The development of more efficient e-waste recycling methods with the help of this research could lead to a more sustainable future.