The field of image and signal processing is witnessing a significant shift towards the adoption of wavelet-based methods, which are being increasingly used to improve the precision and efficiency of various tasks such as segmentation, unmixing, and representation learning. These methods are particularly effective in capturing high-frequency details and preserving semantic context, leading to state-of-the-art performance in several benchmarks. Notable papers in this area include WaveSeg, which proposes a novel decoder architecture that jointly optimizes feature refinement in spatial and wavelet domains, and SWAN, which presents a self-supervised wavelet neural network for hyperspectral image unmixing. Additionally, WaveMAE introduces a masked autoencoding framework that leverages wavelet decomposition to learn scale-aware representations, while T-REGS proposes a regularization framework based on the length of the Minimum Spanning Tree to enhance representation quality. These innovative approaches are advancing the field and opening up new avenues for research and applications.