Human body characteristic parameter prediction method based on semi-supervised learning

Abstract

The invention discloses a human body characteristic parameter prediction method based on semi-supervised learning. The method comprises the following steps of constructing a data set, wherein the dataset comprises a labeled data set based on a real human body and a label-free data set based on a virtual human body; preprocessing the image of the data set; training a semi-supervised model by usingthe data set, and constructing a stable mapping model between the input image and the human body characteristic parameters; and processing the to-be-detected input image by using the semi-supervisedmodel, and predicting to obtain human body characteristic parameters. Only a small amount of real labeled human body data is collected. A large amount of unlabeled human body data is generated by means of the simulator. A stable semi-supervised model mapping model can be established by means of a small amount of labeled human body data, and human body characteristic parameters are accurately predicted.

Description

technical field [0001] The invention relates to the field of human body feature parameter prediction, in particular to a human body feature parameter prediction method based on semi-supervised learning. Background technique [0002] Human body characteristic parameters represented by height, weight, measurements, arm length, etc. reflect the spatial position relationship between human body feature points, and represent human body shape information, and have been widely used in 3D human body reconstruction, virtual fitting and other fields. [0003] With the rapid development of smart phones, the cost of obtaining clear and stable human body images is gradually reduced. Compared with the manual calibration of human body feature points with a huge workload, researchers are constantly trying to build a stable mapping model between human body images and human body feature parameters. . However, the positioning of human body feature points based on image processing is usually ea...

AI Technical Summary

Problems solved by technology

[0005] The labeled human body feature parameter dataset contains a large number of real human body images and corresponding human body feature parameters. The construction workload is heavy and the cost is high. Currently, there is no open source labeled human body feature parameter dataset, so it cannot be used for supervised learning. The model provides enough training sample support

[0006] As another important part of the deep learning field, the unsupervised learning model does not require labels for training samples, and the acquisition cost is significantly reduced. Although it has achieved good classification results in sample classification problems, it cannot be applied to data regression problems such as human body characteristic parameters.

Images