Saliency image generation method and device and server

Abstract

The invention provides a saliency image generation method and device and a server, and the method comprises the steps: obtaining the forgetting degree features of all test users, dividing the test users into a plurality of groups according to the forgetting degree features of the test users, and enabling the forgetting degree features of all groups to be the same; obtaining a clear sample image set and a labeled sample image set of each group, wherein the labeled sample image set corresponds to the clear sample image set; adopting a neural network model to train the labeled sample image set and the clear sample image of each group, and constructing a prediction model of each group; obtaining the forgetting degree feature of the target user, determining the target prediction model accordingto the forgetting degree feature of the target user and the prediction model, and inputting the target image into the target prediction model to generate the saliency image, so that the matching degree of the generated saliency image and the saliency of the actual visual system of the user is improved.

Description

technical field [0001] The present invention relates to the field of computer vision, in particular to a salient image generation method, device and server. Background technique [0002] The salience of the human visual system has the ability to quickly search and locate objects of interest. When people face natural scenes, they can quickly filter out unimportant information according to the salience of the visual system, allowing us to focus more on the senses. area of ​​interest. Therefore, in the application scenario of image push, by predicting the saliency of the visual system of different users, and generating the image to be pushed according to the predicted image saliency prediction result, the user's interest in the image to be pushed is improved. [0003] In the prior art, most saliency detection methods adopt deep learning methods, and are trained according to saliency image samples submitted by users, that is, combining convolution operation with pooling operati...

AI Technical Summary

Problems solved by technology

[0004] However, in the prior art, the user’s forgetfulness of images will lead to the loss of some image feature information in the saliency image samples submitted by users, making the accuracy of the saliency prediction model low, resulting in The saliency mismatch of the user's actual visual system

Images