This week is just the proposal week, so the project specification and scope are being currently defined. Beyond this, I have done a bit of research into the API's and the plausibility of different features I have planned.

For this week I've been exploring different methods of pose detection, and after seeing unsatisfactory accuracy with the preview I am transitioning into having the user record a video which then gets processed.

This week, I got video uploading working, and I was able to extract the frames from the preview. The problem is this was incredibly slow, and it would skip frames. I am currently looking into several different alternatives, including using a codec to quickly extract the frames for my preprocessing pipeline. I also have a pose detector initialized and working in my wrapper.

Finally, I went crazy and did a ton of stuff. I got a video frame preview working, and from there, I pass it to my custom pose detector processor.  It then gets passed into a fragment which passed the relevant information into another analysis class which then figures out stuff like range of motion, reps, angles of body, etc. to give personalized instruction and coaching. I also implemented a login through firebase with google. The user also has a workout planner, where they get a Push-Pull-Legs workout schedule. They can log their sets, and modify the weights which all get saved to their profile. All of this data is backed up on the realtime database from firebase, and can be reloaded on any device. They also can see on the home screen a little recap of their stats, including their total weight lifted, and how much time they've spent as a whole. There is also a settings menu where they can choose which video processing model to use, how much information they want, and even an option to delete their account.