Image weak supervision target detection method based on class agnostic foreground mining

Abstract

The invention provides an image weak supervision target detection method based on class-agnostic foreground mining. The method comprises the steps of generating a foreground attention map through a CNN based on an image to be subjected to target detection; calculating the foreground relative confidence FRC of each candidate frame based on the foreground attention map, and screening out foreground candidate frames according to the FRC of each candidate frame; constructing an instance space graph based on the foreground candidate frames, constructing a tag semantic graph based on tags of the data set, performing graph matching on the instance space graph and the tag semantic graph, and classifying each foreground candidate frame according to a graph matching result; and generating a pseudo supervision frame according to the classification result of each foreground candidate frame, merging the pseudo supervision frame and a spatial neighbor frame of the pseudo supervision frame to obtain a pseudo instance label, and taking the pseudo instance label as a target detection result of the image to be subjected to target detection. According to the method, positioning and classification tasks are separated, so that bidirectional improvement of positioning and classification performance is realized, and weak supervision target detection performance of the image is effectively improved.

Description

technical field [0001] The invention relates to the technical field of target detection, in particular to an image weakly supervised target detection method based on class-agnostic foreground mining. Background technique [0002] Object detection is an important research direction in the field of computer vision and has a wide range of applications in real life. With the development of deep learning, object detection technology has also made great research progress. However, object detection in fully supervised mode requires accurate annotation frames, and the annotation cost is high. On the contrary, the category information of the image is easier to obtain, so people start to study weakly supervised object detection, and complete the object detection task under the condition of only given the image category information. [0003] WSDDN (Weakly Supervised Deep Detection Networks, Weakly Supervised Detection Networks) proposed in 2015 designed weakly supervised target detec...

AI Technical Summary

Problems solved by technology

[0011] Secondly, most of the work does not consider the relationship between the candidate box and the label, and a few works only consider the relationship between the candidate box and the label

Images