I am currently working on a new project aimed at tracking treadmill steps in an app via a front-facing camera. I'm still in the early idea phase and I'm trying to decide if it would be better/more accurate to track a person's steps via something like Pose Detection in OpenCv or via some sort of Object Detection/tracking. My thought process with the latter is that maybe the user could wear something like a red glove that the program can identify and track. When said glove goes up (as a person jogs, for example) a single step is logged. I feel like Pose Detection would be easier to implement and fewer steps, but I have never used the program and I'm not sure how well it would log what would probably be only the front, upper half of a person who is running in place. In contrast, with object tracking, people run/walk/move differently. And if you're walking, you have much less hand movement so the program might struggle to identify the object and movement.
Absolutely any and every suggestion and tip are very much appreciated and needed. Thank you!