A system and method for detecting human body parts

Abstract

The invention discloses a system and method for detecting parts of a human body, comprising: a preliminary screening module for preliminary prediction of a predetermined area of ​​a human body in an input image, and outputs the first boundary range of the predetermined area of ​​a human body; a discrimination and correction module for filtering remove the wrongly predicted area in the first boundary range, correct the correctly predicted area in the first boundary range, and output the second boundary range of the predetermined human body area; also include: an accurate prediction module for Correcting the second boundary range, outputting a third boundary range of the predetermined area of ​​the human body, and predicting predetermined key points in the third boundary range. The technical solution provided by the invention can detect various parts of the human body accurately and in real time on the mobile terminal, and can track the various parts of the human body in real time, so as to meet the market demand.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a system and method for detecting human body parts. Background technique [0002] The detection of various parts of the human body is especially useful in photography and videography. When the user is shooting, after the human body is detected by the photographic camera equipment, the user can set the human body area as the focus and metering center, so that the photos taken will not cause the human body to be blurred due to out-of-focus, and will not be caused by errors in the metering point. As a result, the human body is too bright or too dark; or, composition suggestions are given based on the detected human body position, and even the clothing, gender, age and other characteristics of the person are further detected based on the human body frame. Take the detection of key points of human bones and hands as an example. When key points of bones are detected, interest...

AI Technical Summary

Problems solved by technology

However, these two algorithms need to run on high-performance servers due to their complex models, huge amount of computation, and huge memory usage. Ordinary mobile devices, such as iPhone devices, have a high probability of crashing if the APP occupies more than 500M of memory.

Therefore, the existing systems and algorithms for detecting various parts of the human body cannot be directly used on existing mobile devices

Images