Human Body Feature Point Detection


Background


With the rapid development of "big data" technology and high-performance processors, machine vision and deep neural network technologies have been deeply studied, and the detection rate and accuracy of human body feature points have been greatly improved, which can be used for action recognition, human-computer interaction, and abnormal behavior detection , people tracking and other commercial applications.

image.png


Introduction


Human Body Feature Point Detection Solution is based on Firefly high-performance mini computer, fusing with machine vision and deep neural network technology to detect and locate the key feature points of the human body in front of the camera, including the key feature points of the bones and the key points of the fingers. It accurately detects the key points of the human body, and can be more intelligently applied to the related fields of computer vision.


Kit


Firefly RK3399Pro Mainoard + Single-Lens Camera

image.png

Features


High-performance AI processor

The high-performance AI processor with RK3399Pro and ARM 6-core processor architecture, whose frequency is up to 1.8GHz, has quad-core GPU Mali-T860 MP4 and integrated neural network processor NPU whose computing power is up to 3.0Tops,  compatible with multiple AI frameworks.


image.png


High detection accuracy

Equipped with an HD single-lens camera, it can clearly detect the key feature points of the human body. The frame rate of human bone feature point detection is 15fps/640*480, and the detection accuracy is 95%; the key point detection frame rate of  finger is 6fps/640*480, and the detection accuracy is 92%.

image.png

Applications


It can be used in indoor/outdoor vending machines, unmanned supermarkets, new retails and higher level entertainment.

image.png


Contact us at sales@t-firefly.com for more solutions