Faster R-CNN based substation scene image detection method
Wei Dong Zhu, Guoping Mu, Jun Li · 2021
In recent years, salient target detection has become one of the hot research issues in the field of computer vision. This paper designs a salient target detection method based on Faster R-CNN in a scene with relatively complex background information. First, perform multi-scale superpixel segmentation on the image, and use Faster R-CNN to perform target detection on the image, and perform saliency screening on superpixels according to the characteristics of similarity. After obtaining the initial target location feature, perform saliency detection and optimization, and finally use the primitive The cellular automata method fuses multi-scale superpixel saliency maps. By conducting experiments on specific types of data sets and comparing and analyzing existing typical saliency detection, it is verified that the method in this paper can improve the accuracy of saliency detection in images with complex background.