Picture category identification method and device

Abstract

The embodiment of the invention provides a picture category identification method and device, and the method comprises the steps: inputting a sample picture into a pre-selected model, and carrying outthe prediction, so as to obtain the prediction category and corresponding prediction probability of the sample picture; calculating a loss value according to the prediction probability and the expected prediction probability of the sample picture by adopting a preset first weight for a correct sample and a preset second weight for an error sample, with the first weight being smaller than the second weight; and adjusting parameters of the pre-selected model according to the loss value to continue training until the total loss value is smaller than a preset loss value threshold value, and ending the training. In the above method, when the loss value of image classification model training is calculated, the first weight corresponding to the correct sample is set to be smaller than the secondweight of the error sample, so that the training process can pay more attention to the sample which is difficult to pay more attention to, thereby accelerating the reduction speed of the loss value,and further accelerating the convergence speed of the model at lower cost.

Description

technical field [0001] The present invention relates to the field of deep learning, in particular to a method and device for identifying picture categories. Background technique [0002] At present, deep learning is the focus of research in the field of artificial intelligence, and a large number of scholars and researchers have devoted themselves to it, promoting its rapid development. Although deep learning has made great achievements, it still faces many problems. Especially in the field of image classification, a large amount of training data and multi-possibility classification probabilities generate more intermediate data during the training process, which means that image classification model training requires more training storage space and time; training an image Classification models often take days or even months, so accelerating the training process and saving time and cost is an important research direction at present. [0003] For accelerated training, in the...

AI Technical Summary

Problems solved by technology

[0004] In view of this, the embodiment of the present invention proposes a method and device for identifying image categories to efficiently solve the problem of slow convergence of existing image classification domain models

Images