Image recognition method and device and computer readable storage medium

Abstract

The invention relates to image processing, particularly provides an image recognition method and device and a computer readable storage medium, and aims to solve the problem of how to perform accurate image recognition on target objects in different image states. In order to achieve the purpose, the method comprises the steps that image samples are classified according to the image state of a target object in an image to obtain image sample sets of different image states, a recognition model is trained through the image sample sets with labels, and an initial image recognition model is obtained; classification adversarial learning training is carried out on the initial image recognition model by adopting the image sample set with the labels and an image sample set without a label, and feature regression training is carried out on the initial image recognition model according to an image feature distance between the image sample sets with the labels and each image sample set without the label to obtain a final image recognition model. Accurate image recognition can be carried out on the target objects in different image states by adopting the final image recognition model.

Description

technical field [0001] The present invention relates to the technical field of image processing, and specifically provides an image recognition method, device, and computer-readable storage medium. Background technique [0002] Image recognition refers to the use of computer equipment to analyze and process images to identify target objects in images, such as face recognition or object recognition. At present, the conventional image recognition method mainly uses the image samples marked with sample labels to train the image recognition model, and then uses the trained image recognition model to perform image recognition on the image. In order to improve the recognition accuracy of the image recognition model, it is often necessary to use a large number of image samples for model training. However, if the image samples are labeled one by one, it is not only time-consuming and labor-intensive, but also extremely error-prone, which significantly reduces the accuracy of the im...

AI Technical Summary

Problems solved by technology

However, if the image samples are labeled one by one, it is not only time-consuming and labor-intensive, but also extremely error-prone, which significantly reduces the accuracy of the image recognition method.

In addition, in practical applications, the target object in the image to be recognized may be in various image states such as clear, blurred, and occluded. If the image samples of each image state are labeled with sample labels one by one, it will be more time-consuming and laborious.

Therefore, the current conventional image recognition methods often only perform image recognition on target objects in one image state, and cannot perform image recognition on target objects in different image states at the same time.

Images