Clothing identification method, system and terminal based on human body key node prediction

Abstract

The invention provides a clothing identification method based on human body key node prediction. The clothing identification method comprises the steps of data collection, wherein an original image containing a human body posture is acquired; human body posture detection, wherein a human body posture is detected in the original image, and human body posture key node information required in clothing identification is obtained; and clothing identification, wherein a target area containing clothes to be identified is determined according to the human body posture key node information, and the target area is processed by adopting a convolutional neural network to obtain a clothes identification result. Meanwhile, the invention provides a clothing identification device and system based on humanbody key node prediction and a terminal. According to the invention, the intelligence of personnel and clothing compliance supervision is effectively improved, and the working intensity of supervisors is reduced.

Description

Technical field [0001] The present invention relates to the technical field of image processing, in particular to a clothing recognition method, system and terminal based on the prediction of key human body nodes. Background technique [0002] In many industries at present, there are certain requirements for dressing, such as the dressing of employees in the food industry and the dressing of medical staff. By intelligently identifying surveillance video, it can replace supervisors to supervise various industries such as the food industry, and effectively improve the efficiency of safety supervision. [0003] In order to identify the clothing category of a person, the video image is preprocessed by means of mean filtering, light compensation, and normalization, and then the difference method is used to separate the foreground and background, and then methods such as binarization, corrosion, and expansion are used. Morphological difference processing is performed on the image, and t...

AI Technical Summary

Problems solved by technology

[0004] At present, the clothing compliance inspection of practitioners on the market needs to design a set of detection methods for the detected head, torso, and face, and then design a set of corresponding clothing and mask recognition methods to determine whether they are wearing appropriate clothing, hats, and masks. Increased the complexity of the entire detection method

In addition, this method requires R&D personnel to select appropriate features for a specific application scenario, resulting in a relatively stable effect of the method in identifying a specific color and style of clothing and masks in this specific scenario, but different in another scenario. The detection and recognition effect of the color and style of hats and masks is likely to be poor, and it cannot be deployed on a large scale in different application scenarios

[0005] At present, there is no description or report of the similar technology of the present invention, and no similar data at home and abroad have been collected yet.

Images