Multi-scale information enhanced binocular convolutional neural network saliency image detection method

Abstract

The invention discloses a multi-scale information enhanced binocular convolutional neural network saliency image detection method, and relates to the technical field of neural networks. The method comprises the steps of constructing an end-to-end convolutional neural network binary classification training model, inputting original scene image data in a training set into the convolutional neural network binary classification training model for training, and obtaining a final saliency prediction image; and repeatedly calculating a loss function value between the final saliency prediction image and the real saliency detection image set to obtain a trained convolutional neural network binary classification training model, and inputting a scene image to be subjected to saliency detection into the convolutional neural network binary classification training model to obtain a predicted saliency detection image. The saliency detection image generated by using the method is clear in boundary andcomplete in structure, and the problems of structural loss and incomplete details of the saliency detection image generated by using the existing method are solved.

Description

technical field [0001] The invention relates to the technical field of neural networks, in particular to a multi-scale information enhanced binocular convolutional neural network saliency image detection method. Background technique [0002] In recent years, due to the rapid increase of computer hardware capabilities and the rise of deep neural networks, many traditional computer vision tasks have been quickly redefined. With the help of deep learning tools, many visual task indicators have been quickly refreshed, including saliency detection. Saliency detection is a detection method based on the attention mechanism of the human brain. When a person sees a scene, humans automatically prioritize some areas of interest, while temporarily ignoring uninteresting areas, and the areas of interest that humans screen through their unique attention mechanism are called salient region, and saliency detection is a task of detecting salient regions through existing means, which is simi...

AI Technical Summary

Problems solved by technology

[0005] In view of this, the present invention provides a multi-scale information-enhanced binocular convolutional neural network saliency image detection method to solve the problem of insufficient utilization of feature information proposed in the background technology to effectively obtain saliency images

Images