A Method of Target Classification and Localization Based on Network Supervision

Abstract

The invention provides a target classification and positioning method based on network supervision, comprising the following steps: automatically obtain a large amount of network image data from a search engine according to the category of the target to be tested; filter and remove noise images to form a training sample set; initially construct classification and Positioning network: Input the samples in the training sample set into the initially constructed classification and positioning network for feature extraction, classify the features, and obtain the position information of the target object, and implement the training of the classification and positioning network. In the present invention, the end-to-end fine classification and positioning method based on network supervision uses a large number of easily obtained network images as a training set, completely removes manual annotation, only uses image-level labels, designs an efficient convolution network, and integrates the global average Algorithms such as pooling and class activation maps enable the present invention to outperform weakly supervised learning methods in fine classification tasks and positioning tasks.

Description

technical field [0001] The invention relates to the fields of computer vision and image processing, and in particular to a network-supervised target classification and positioning method that can be used in intelligent automatic recognition and other directions. The method can be widely used in the field of automatic recognition of mobile phones. Background technique [0002] The target positioning and detection tasks under fully supervised and weakly supervised learning have developed rapidly in recent years, and the most advanced performance is constantly being refreshed. How to further improve the performance? Obviously, designing a deeper network or using more training data are two directions that researchers are exploring. In fact, designing a deeper network is bound to expand data, so how to provide more data to the network is a key issue to be studied. [0003] With a large amount of online visual data, the Internet and social media have become the most important dat...

AI Technical Summary

Problems solved by technology

[0007] The above network supervised learning methods have more or less added human intervention or auxiliary labeling, and the performance is still not comparable to weakly supervised learning

Therefore, there are still many problems in the network supervised learning method, and there is a lot of room for improvement.

Images