Image processing method and storage medium

Abstract

The invention provides an image processing method. The image processing method comprises the steps of obtaining at least one image of a to-be-detected target; generating a feature heat map based on the image; inputting the image and the feature heat map into a deep learning model at the same time to obtain a training feature value; and obtaining a prediction result from the deep learning model according to the training feature value. The invention further provides a computer readable storage medium. According to the method, the accuracy of target judgment in the image can be improved.

Description

technical field [0001] The invention relates to the technical field of image recognition, in particular to an image processing method and a storage medium. Background technique [0002] The use of deep learning convolutional networks to discriminate images is the current mainstream, and it has excellent performance in discriminating images. It can also be applied to target detection in images, but the deep learning convolutional network also has certain limitations in discriminating targets. For example, it is difficult for developers to explain the model generated by the convolutional network, and they can only predict its quality from the discrimination results of the model. Therefore, it is not easy to carry out targeted control and improvement on the network. On the other hand, convolutional networks tend to ignore the correlation between the local and the whole. For example, it cannot extract features such as the "position", "size" and "direction" of the target in th...

AI Technical Summary

Problems solved by technology

It can also be applied to the target detection of images, but the deep learning convolution network also has certain limitations in discriminating targets.

For example, it is difficult for developers to explain the model generated by the convolutional network, and they can only predict its quality from the discrimination results of the model

Therefore, it is not easy to carry out targeted control and improvement on the network

On the other hand, the convolutional network tends to ignore the correlation between the local and the whole

For example, it cannot extract features such as the "position", "size" and "direction" of the target in the image

However, in some cases, these factors are an important basis for target judgment, and the lack of judgment of these factors will affect the accuracy of the model for target discrimination.

Images