Scene classification method and device based on self-supervision mechanism and regional suggestion network

Abstract

The invention discloses a scene classification method and device based on a self-supervision mechanism and a regional suggestion network, and belongs to the field of image processing, and the method comprises the steps: obtaining a standard data set of scene classification, abiding by the division rules of different standard data sets, and dividing the data set into a training set and a test set according to a label file; constructing a base network used for extracting features, a region suggestion network used for extracting a local discriminable information region and a joint network used for combining features of different scales; performing optimization design on a loss function of the network by adopting a self-supervision mechanism; alternately training the parameters of the base network, the regional suggestion network and the joint network by using the training set pictures; and inputting the test image into a trained network to obtain a scene category of the image. According to the method, no extra local area labeling information is needed, and only the scene category hierarchical label of the image is needed, so that the network can perform end-to-end learning for different data sets.

Description

technical field [0001] The invention belongs to the field of image processing, and more specifically relates to a scene classification method and device based on a self-supervision mechanism and a region suggestion network. Background technique [0002] In the past two decades, the rapid development of information and Internet technology has led to explosive growth of various types of data. Image data, as an important part of multimedia, has also surged with the promotion of various social networking sites and software. Scene classification has also received more and more attention in recent years, and has a wide range of applications in the fields of automatic driving, image retrieval, and drone flight, and how to accurately classify scenes has become a challenging problem. . [0003] Traditional scene classification methods mainly use hand-designed features to describe images, and then use various supervised classification algorithms to classify the features. Such featur...

AI Technical Summary

Problems solved by technology

[0007] Aiming at the above defects or improvement needs of the prior art, the present invention proposes a scene classification method and device based on a self-supervision mechanism and a region suggestion network, thereby solving the problem that the existing scene image extracts local features that require additional frame labeling information, resulting in operation difficult technical issues

Images