Multi-label image classification method, device and electronic equipment

Abstract

The present invention provides a multi-label image classification method, device and electronic equipment. On the one hand, the pooling layer and the fully connected layer are used to classify the first feature image to obtain the first label classification prediction result. On the other hand, according to the parameters of the fully connected layer and the first feature image, the first feature image is subjected to feature filtering to obtain the second feature image; wherein the parameters of the fully connected layer and the parameters of the convolutional layer are optimized based on a metric learning algorithm; Then pooling is performed on the second feature image to obtain the second label classification prediction result. Finally, the first label classification prediction result and the second label classification prediction result are considered comprehensively to obtain the target label classification prediction result. This method performs label classification from two aspects, and corrects the first label classification prediction result based on the second label classification prediction result obtained from the second feature map, reduces the number of label combinations, and assists in improving the recognition accuracy of multi-label images.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to a multi-label image classification method, device and electronic equipment. Background technique [0002] Multi-label image classification is a very important research topic in computer vision. Because the pictures taken in the real scene always contain multiple objects, so the image contains multiple labels, and the number of combinations of classification results is exponentially higher than that of a single label. Therefore, compared with the single-label image classification problem, the multi-label image classification problem is more difficult, the accuracy is lower, and it has more research significance. [0003] Most of the traditional methods use graphs to model the relationship between labels, so as to artificially impose constraints on the final predicted results in order to reduce the number of classification results. Such a method depends very much...

AI Technical Summary

Problems solved by technology

Therefore, there is still no good solution to the problem of how to use the relationship between labels in a real sense to improve the accuracy of multi-label image recognition.

Images