Image processing method and device, electronic equipment and computer readable storage medium

Abstract

The invention provides an image processing method and device for processing an input image by using a convolutional neural network system, and a computer readable storage medium. A neural network system includes an input layer, an intermediate layer, and an output layer, the image processing method including: receiving an input image via the input layer, and extracting an initial feature map of the input image; further extracting image features from the initial feature map through the intermediate layer to obtain an intermediate layer feature map; and outputting, via the output layer, a processing result for the input image based on the intermediate layer feature map, wherein the intermediate layer includes at least two branches that generate intermediate layer feature maps having different scales and / or different regions corresponding to the initial feature map.

Description

technical field [0001] The present disclosure relates to the field of image processing, and more particularly, the present disclosure relates to an image processing method and device for processing an input image using a convolutional neural network system, and a computer-readable storage medium. Background technique [0002] Neural networks are a tool for large-scale, multi-parameter optimization. Relying on a large amount of training data, the neural network can learn hidden features that are difficult to summarize in the data, thereby completing multiple complex tasks, such as face detection, image semantic segmentation, object detection, action tracking, natural language translation, etc. Neural networks have been widely used in the artificial intelligence community. [0003] When neural networks are used in the field of image processing, as an important supplement to face recognition technology, person re-identification (Re-ID) is a processing task to determine whether...

AI Technical Summary

Problems solved by technology

Although the processing method based on the basic model is relatively simple, it often cannot effectively complete the pedestrian re-identification task because the extracted image features are too rough.

The processing method based on the attention mechanism model has high requirements for the design of the network model itself, and an improper network structure may cause the attention mechanism to be invalid

However, once the preprocessing model is wrong in the processing method based on the prior knowledge model, it will cause significant interference to the subsequent feature learning.

Images