Sensors | Open Access Journal | MDPI Recent advances in deep learning (DL) have introduced diverse architectures, including convolutional neural networks, residual learning, encoder–decoder representations, dual-branch structures, and contrastive learning, to capture specific sensor artifacts