Construction method and application of deep convolutional neural network with resolution adaptability

Abstract

The invention provides a construction method of a deep convolutional neural network with resolution adaptability. The method is used for constructing a deep convolutional neural network model capableof adapting to scale features of target images with various resolutions, and is characterized by comprising the following steps of S1, setting a plurality of target scales according to the target images to form scale levels, and setting a target retrieval step length according to the scale levels; s2, obtaining a training image, and carrying out the standardization of the training image accordingto the order of the size of the training image, thereby obtaining a standardized training image; step S3, designing a deep convolutional neural network model for adapting to multi-resolution input andmulti-scale target detection, training the deep convolutional neural network model through the standardized training image so as to obtain an executable deep convolutional neural network model, wherein the deep convolutional neural network model in the step S3 comprises a feature map extraction part, a step-by-step down-sampling part, a branch convolution operation part and a prediction output part.

Description

technical field [0001] The invention belongs to the field of digital image processing and deep learning, and relates to an algorithm model design of a deep convolutional neural network, in particular to a deep convolutional neural network with resolution adaptability. Background technique [0002] Digital image analysis technology plays an important role in today's society, of which image object detection technology is an important part. At present, the development of target detection technology has gradually abandoned the manual design algorithm scheme of traditional digital image processing, and turned to deep learning, represented by Convolutional Neural Network (CNN), to achieve high-accuracy target detection results. . The deep learning revolution broke out from 2011 to 2012. The deep learning revolution made computer vision reach a practical level in many application fields and gave birth to a large number of applications in the industry. The most important reason is...

AI Technical Summary

Problems solved by technology

This practice is a huge limitation to the practical application of the algorithm

[0006]The existence of this limitation inevitably results in zooming, stretching, cutting and other operations on the image, resulting in the deformation of objects in the image and the loss of a large amount of pixel information

For small-resolution images, enlarging the resolution inserts redundant information and increases additional computational burden; for large-resolution images, shrinking the image wastes valuable image information, resulting in a decrease in accuracy

In the process of the same resolution, the aspect ratio of the image will also be adjusted, so it will also affect the geometric shape of the target in the image, which will reduce the detection effect of the algorithm, and the image with a large difference from the uniform shape cannot even apply the algorithm

Images