An efficient algorithm for detecting small defects in high resolution cloth images

Abstract

The invention relates to an efficient inspection algorithm for small defects in high-resolution cloth images, which comprises the following steps: (1) collecting images through a camera, and labelImgtool images are used for labeling; 2) dividing that processed image into a training set and a t set, wherein the training set is used for training the inspection model, and adopting the t set to evaluating the performance of the inspection model; (3) inputting the training set image and the corresponding category information position information into the improved se- Resnext101 test model, training test model; (4) After training, adopting the inspection model to process the images in the test set to obtain the approximate position and the corresponding category of the defects. The method of the invention can realize the processing of the multi-scale characteristic map to the single-resolution input image, thereby processing the multi-scale image block, so as to adapt to a plurality of defects of different sizes, and greatly improve the detection accuracy and speed; and the method of the invention can realize the processing of the multi-scale characteristic map to the single-resolutioninput image. At that same time, the algorithm can obtain the approximate position of the defect on the image classification frame and deal with the situation that there are many defects in the image.

Description

technical field [0001] The invention relates to the field of image classification, in particular to an efficient inspection algorithm for small defects in large-resolution cloth images. Background technique [0002] The previous image classification mainly used traditional machine learning methods, which are usually divided into two parts: feature extraction-based methods and template-matching-based methods, and feature extraction-based methods mainly include statistics-based methods, spectrum-based methods, and texture-based methods. Model methods, learning-based methods, and structure-based methods. These methods all need to select features artificially, and the generalization is not strong at the same time. [0003] With the development of deep learning, especially the application of convolutional neural network in image classification, image detection and image segmentation, the effect achieved is unmatched by traditional algorithms in the past. However, the classifica...

AI Technical Summary

Problems solved by technology

However, the classification algorithm based on deep learning requires that the target area in the training data set accounts for a large proportion of the image area in order to have a better effect. If the image is a high-resolution image but the classification target is small, it is possible that the classification target only takes up If it is less than 1% of the total area of ​​the image, the accuracy rate is very low if the traditional deep learning-based classification algorithm is used directly for image classification

At the same time, if there are multiple categories of targets in the image, we cannot obtain the category information of these targets

Images