Until now, Vision’s person box detection (DetectHumanRectanglesRequest) could only get the upper body box.
From the 2021 update, it has become possible to detect boxes throughout the body.
Use the new VNDetectHumanRectanglesRequestRevision2.
And set .upperBodyOnly property false.
upperBodyOnly = false
upperBodyOnly = true
thoughts
We don’t have to bother to use a machine learning model to take the whole body.
I wondered if there were cases where it was necessary to detect only the upper body, but in cases where only the upper body is shown in the frame as in this video, the number of detections is higher for only the upper body.
Whole body
Upper Body Only
I’m a freelance engineer.
Work consultation
Please feel free to contact us with a brief development description.
rockyshikoku@gmail.com
I am making an app that uses Core ML and ARKit.
We send machine learning / AR related information.