Multi-label image deep learning classification method and equipment

Abstract

The invention relates to a multi-label learning technology in the field of machine learning, and relates to a multi-label image deep learning classification method and equipment. The method comprisesthe following steps: acquiring a label relation graph; acquiring mapping of all types of labels and mapping of all label groups according to the label relation graph; constructing a deep convolutionalneural network and carrying out image general feature extraction; selecting feature maps of different layers of the convolutional neural network, and mapping the feature maps to a label and label group mapping dimension through a mapping function; calculating a conformity score and a normalization score of the label and the label group at the current pixel point position for all the pixel pointsin the selected feature maps; acquiring a final label related semantic feature and a final label group related semantic feature; and performing label prediction. According to the method, the label relationship is effectively utilized, richer image general features and label relationship features are learned, and a multi-label classification task is better carried out.

Description

technical field [0001] The present invention relates to multi-label learning technology in the field of machine learning, relates to graph embedding learning and classification technology in deep multi-label learning, and in particular to a multi-label image deep learning classification method and equipment. Background technique [0002] In the era of big data, multi-label images are becoming more and more complex. The complexity of multi-label images is not only reflected in the increase in the number of labels in the image, but also in the increasingly complex distribution of different labels in multi-label images. In order to solve the classification problem of multi-label images, in addition to using the features of the image itself such as outline, shape, color, etc. for label classification, it is also possible to model the label relationship by combining the interrelationships between labels in multi-label learning. [0003] The current deep learning of multi-label im...

AI Technical Summary

Problems solved by technology

Although these methods take advantage of the similarity between tags, they do not take into account the higher-order tag similarity

At the same time, there is no further constraint on the label relationship by grouping the labels according to their own attributes.

[0006] In summary, the deep learning classification of multi-label images uses the convolutional neural network in the deep learning method as the feature extractor, and the use of different convolutional neural network models and the features of different layers of the same convolutional neural network will affect the final classification results. , there is currently no effective method for the integration of features from different layers of the same convolutional neural network model

Although the extraction of label relationship features has different methods, they do not fully consider the multi-level label similarity.

Images