An Improved YOLOv7 Model Based on Visual Attention Fusion: Application to the Recognition of Bouncing Locks in Substation Power Cabinets

With the continuous progress of intelligent power system technology, in order to meet needs substation operation and maintenance, a target detection algorithm is applied identify status equipment switches. YOLOv7, as latest achievement YOLO (You Only Look Once) series algorithms, has good speed accuracy tasks. However, when generalized network specific scenario, its advantages are not obvious due high weight poor portability. In this paper, an improved GF-YOLOv7 model proposed apply recognition bounce locks substation. The MobileViT module used improve feature extraction ability backbone network. Referring CBAM attention mechanism, channel spatial design more lightweight fusion experimental results test set show that can significantly reduce on basis small reduction speed, reaches 97.8%, which task locks.