Lightweight visual backbone network with enhanced comprehensive strength through context-aware dual attention mechanism
Abstract: Highlights•A context-aware Decoupled Fully Connected (CADFC) attention mechanism is proposed.•A new CADFC bottleneck block aggregates two branches: depthwise and CADFC branches.•A CADFC MobileNet is proposed, optimizing MobileNetV2 for image classification tasks.
External IDs:dblp:journals/ijon/XueHHCWCL25
Loading