Lightweight human ear recognition method and system based on attention mechanism and feature fusion

The invention discloses a lightweight human ear recognition method and system based on an attention mechanism and feature fusion, and the method comprises the steps: improving a YOLOv5s algorithm architecture, constructing a lightweight network, setting attention modules at the joints of a backbone network and a neck network of the constructed lightweight network, and obtaining a human ear recognition model; human ear data is acquired, human ear recognition is performed through a constructed human ear recognition model, attention weighting on different dimensions is performed on a target in the feature map, feature fusion is performed through a neck network of the human ear recognition model, and human ear recognition is performed through an output network. The feature extraction of the backbone network of YOLOv5s and/or the feature fusion of the neck network are/is lightened at the same time, the real-time performance of the method can be greatly improved, meanwhile, a CBAM attention mechanism and cross-regi