Had this idea in mind for awhile and while there are some apps currently I’ve always thought about if this was possible to some extent – an app for tracking/perfecting your squat form, which an MVP version might just be tracking knee valgus – in which while there are some exceptions (max lifting/clean and jerk), generally when one experiences knees caving in during a squat – knee valgus – it’s recommended to address for whatever actual root issue it may be – whether it’s a case of imbalances in the hips – adductors (inner thigh muscles) overpowering abductors (outer glutes) or ankles aren’t mobile enough. A really tuned-in person, or someone using a mirror/video or partner can notice this it might be able to address it – but what about having a program that flags this for you so you can address it immediately? Maybe eventually tuned to recording enough that it can spot patterns with enough data to have some algorithms with recommendations?
Knee valgus might be easier to spot with pose detection with a front-facing camera, but for a more thorough squat form there would be various things to consider – Not all bodies are the same, as we can have different proportions of femur/tibia/torso, hip socket depth, ankle mobility which determine how upright we can be, ideal foot positioning. Ideally maybe some inputted data of either manual measurements, or a trifecta/quadfecta of cameras (front, back, 2 side angles?) would be ideal to be able to assess other weaknesses such as butt-wink or shifting, depth problems due to ankle mobility/foot positioning, excessive forward-lean to name a few.