Garment image automatic segmentation method and mechanism based on deep learning

Abstract

The invention belongs to the technical field of clothing image automatic segmentation, and particularly relates to a clothing image automatic segmentation method and mechanism based on deep learning. The clothing image automatic segmentation method based on deep learning comprises the following steps: collecting a certain amount of clothing image data, training a convolutional attitude machine network, and outputting a clothing key point model; inputting a to-be-processed image, obtaining a clothing key point model, and forming a GrabCut initial rectangular frame on the to-be-processed image; and carrying out segmentation processing on the to-be-processed image forming the GrabCut initial rectangular frame. According to the method, the key point information is output by using the convolutional attitude machine network, and the horizontal coordinate, the vertical coordinate and the circle radius of the picture where the key point information is located are obtained, so that the GrabCut initial rectangular frame is formed, and the background extraction effect is improved. According to the method, the problem of manual interaction is solved, the purpose of automatic image extraction is achieved, an excellent segmentation effect is achieved for images with complex backgrounds, and the method has high practicability for automatic segmentation processing of large-batch garment images.

Description

technical field [0001] The invention belongs to the technical field of automatic clothing image segmentation, and in particular relates to an automatic clothing image segmentation method and mechanism based on deep learning. Background technique [0002] Image segmentation technology, as part of image preprocessing, has been widely used in fabric defect inspection, fabric pattern contour extraction, clothing contour extraction, clothing image contour extraction, clothing image retrieval, etc., but in practical applications, there are still many problems. question. [0003] The existing schemes generally use the convolutional neural network method to segment the clothing image. This method can accurately locate the position of the clothing in the image, but the image segmentation edge effect extraction is general, and there are certain requirements for the quality of the training set. [0004] The GrabCut algorithm is an image segmentation method based on graph theory. Altho...

AI Technical Summary

Problems solved by technology

[0004] The GrabCut algorithm is an image segmentation method based on graph theory. Although it has the advantages of high precision, it also has disadvantages: it requires a small amount of manual interaction by the user, and cannot automatically process batch images; the extraction effect of single background and complex background is quite different.

Images