I dedicated additional time to refining the comparison algorithm in our project, which focuses on capturing single image inputs with local reference posture. During our Friday meeting, the rest of the team conducted a code review of the algorithm and highlighted several areas that needed optimization. This was necessary because some parts of the code had originally been hardcoded for testing purposes, Hongzhe helped me pointed out some of the issues in my code. After addressing and enhancing the script based on our team members’ feedback, we managed to align it with most of the initial design goals. This included functionality for scoring, generating joint angles as outputs, and integrating our voice engine. I worked closely with Hongzhe this week in developing scripts and practiced some unit testing in different environments based on his output from the Openpose script.
As for the voice module implementation, we have not yet settled on a specific approach. I will continue collaborating with the rest of the team to determine how to incorporate the voice component into our program. I am currently attempting to write a separate script to call on and test generated voice files, but I would need to do some testing before integrating that part possibly on Sunday or during the next week. If the separate script could not be integrated in time, we will do a manual approach for the interim demo on this part and just showcase that the generated .wav files could be played.