Intelligent safety monitoring of weed defects in capacitor bank fence based on fusion data

This paper designs a model combining the spatial location features of point cloud data and the color and texture features of image data is designed to detect weed defects in the fence of capacitor bank. The dense feature fusion module and the adaptive fusion module improve the performance indicators of the model. Compared with the image target detection method, the detection accuracy of the visual fusion model is improved from 0.79 to 0.88, which effectively improves the detection ability of the algorithm.