Abstract:To assist operators in accurately determining the status of equipment switches,more and more intelligent inspection robots are introduced into distribution station buildings for equipment status monitoring,This article proposes a Single Shot MultiBox Detector image recognition algorithm based on the hardware of intelligent inspection robots participating in collecting image data,to achieve recognition of two-phase handle switches.Improving the algorithm by embedding the Convolutional Block Attention Module into the SSD algorithm,selects effective features from channel and spatial dimensions,suppresses irrelevant features,and improves the recognition accuracy of the model.The experimental results show that the AP values for detecting switch status targets are all above 98.66%, the mAP of the improved algorithm is 99.10%,which can better identify switch status compared to traditional SSD algorithms and improve the intelligence level of substations.