I need a way to Compare Body Movements of a saved video (the video could have been saved after the real time tracking with ML kit/ MediaPipe, or just being a simple uploaded video you had in your gallery) against another video that is being recorded in real time using the camera and giving real time feedback to the user of how they're performing the body movement compared to the saved video.
I wanted to create this for my native Android Project, but I'm kinda a beginner in the AI world, so i didn't know which was the best approach in my case. it is possible to just use ML kit?, if not should i use Mediapipe and create everything in python and then use Chaquopy for integrating it in my project? Should i create everything just in Kotlin and integrating mediapipe there? The idea is to make the project have the best performance and do it in the simplest way.
For what i have been reading, i could use FastDTW for Temporal Alignment and Cosine distance as the metric for comparing the similarity of poses between the two videos, but still not sure how to combine everything together to do it, i would really appreciate your help c: