FormCheck

Home page
Pro Comparison #1
Video Vault
Pro Comparison #2
Pro Comparison #3

Have you tried learning something, but you never had the right tools to help you? Lucky for you, we made FormCheck. It's a computer vision tool that uses a video of you and another professional of your choice to analyze the differences in your forms.

We've gone through the pain of learning a new sport without having anyone to teach us form, technique, or otherwise how to get better. Especially for many college students with no prior athletic background, getting consistent coaching is expensive and not to mention hard to come by. We've longed for a tool that would compare our forms to the best of the best.

We built the program using openCV and the MediaPipe Pose Estimation AI to give a geometry of your body. We compare this to a professional in your field of interest that you supply combined with (in a perfect world) pre-uploaded videos that match your tract.

The minor inconveniences that we faced were UI issues and skeletal imprint of the body difficulty overlaying two skeletal bodies. There were also several issues with identifying the correct POI especially in situations with more than one individual. This was solved by letting the users choose which person to track when uploading the clips, rendering the skeletal imprint to compare the POI and user's form with utmost accuracy.

We started with learning OpenCV which was the foundation of the whole project. We haven't used CV before, so implementing it in such a complex manner with a project so close to our interests was cool in its own right. We took it a step further by going much more into the unknown by using a new IDE (Antigravity). We used Python as our main language as it's a strong yet flexible base that led us to learn PyQT and QSS. This helped our front-end look modern yet customizable.

Currently, it's almost all locally run, but in the future we want to have mostly everything cloud-based as opposed to locally accessible (especially the database) . We want to make the pose estimation more accurate and expand into multi-disciplinary motor skills. Especially overlooked disciplines like dancing, typing, or playing an instrument. These are all equally valid in our eyes, and we'd like to expand the opportunities and service they to their community.