CA-Stream: Attention-based pooling for interpretable image recognition

arXiv:2404.14996v1 Announce Type: new
Abstract: Explanations obtained from transformer-based architectures in the form of raw attention, can be seen as a class-agnostic saliency map. Additionally, attention-based pooling serves as a form of masking the in feature space. Motivated by this observation, we design an attention-based pooling mechanism intended to replace Global Average Pooling (GAP) at inference. This mechanism, called Cross-Attention Stream (CA-Stream), comprises a stream of cross attention blocks interacting with features at different network depths. CA-Stream enhances interpretability in …

abstract architectures arxiv attention class cs.cv design feature form global image image recognition map masking observation pooling raw recognition space transformer type