Clothes color recognition method, device and system based on semantic segmentation

Abstract

The invention provides a clothes color recognition method based on semantic segmentation, and the method comprises the steps: collecting a certain number of portrait pictures with different parametersin different scenes, and marking the collected portrait pictures; carrying out random parameter transformation processing on the marked portrait picture to generate an initial sample set; creating aclothing region extraction model based on a JPPNet network, and extracting a local picture of each clothing region contained in each portrait picture; carrying out background transformation and colorclassification labeling on the extracted local picture; performing size unification and random parameter transformation processing on the marked local pictures to generate a training sample set; and creating a clothing color recognition model based on the classifier, and importing the training sample set into the clothing color recognition model to train the clothing color recognition model. According to the method, higher recognition success rate and recognition accuracy can be achieved, and particularly, the improvement amplitude of the recognition success rate and the recognition accuracy is higher under the complex conditions that only a half figure exists in a picture, and shielding exists in the picture.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to a clothing color recognition method, device and system based on semantic segmentation. Background technique [0002] The recognition of clothing color generally classifies the upper and lower clothing colors directly by inputting the whole portrait into the classifier, but this method is not suitable for incomplete portraits caused by shooting angles, occlusions, etc., or complex postures of portraits. difficult problem to judge. [0003] In addition to directly using the classifier to judge, the industry also has multi-label classification based on the graph neural network. The reason for this is that the relationship between the color of the upper garment, the color of the lower garment, and other attributes of the human body can be used. There is no obvious connection between attributes such as clothing style and clothing length, which leads to a certain degr...

AI Technical Summary

Problems solved by technology

[0006] However, in practical applications, these schemes are difficult to adapt to the identification of complex situations. For example, in the aforementioned patent, the clothing area must be positioned based on the coordinate point information of the clothing. When part of the coordinate point information of a certain clothing area is blocked or When the loss is caused by shooting angle and other reasons, it is necessary to fit the lost coordinate points. In addition to the slow recognition speed due to the large amount of calculation in the fitting process, the low precision of the fitted coordinate points also makes the actual application The final recognition effect is greatly reduced

[0007] To sum up, the above-mentioned technology still has many problems in color judgment, such as misjudgment, missed judgment, etc., resulting in inaccurate color judgment in many cases, and it is difficult to achieve practical purposes.

Images