Inspiration - We were tired of not being able to easily improve our workout form with compound movements (think bench press, squats, deadlifts) at the gym. Personal trainers are expensive and you don't always have a gym buddy. So we decided to build an app to fix this.
What it does - Our application analyzes the user's body using Pose-estimation technology and is able to accurately count reps for different workouts. Such as for Shoulder Press, Squats, and barbell rows.
How we built it - We built this as a web-based application.
Essentially, we created a full-stack web application that incorporates computer vision (OpenCV), and 3d pose estimation (MediaPipe) to create software that can perceive, understand, and estimate your movements. We developed our front-end styling and deployment using Figma, notion, HTML, CSS, ajax, and flask integration. We are hosting our website locally, using flask for our backend, and plan to use Firebase for our database. Finally, to find, and understand human movement, we used OpenCV and eventually YoloV4 to classify workout poses. MediaPipe will be used to calculate their estimated pose.
Challenges we ran into - Not having enough time for a lot of things. We developed the app proof-of-concept on Figma but weren't able to deploy in time, so we stuck with the web application for now. Another issue we ran into was deploying to a web server, instead of being hosted locally.
Accomplishments that we're proud of - We're incredibly proud of our team for coming together and leveraging their various skills for different aspects of the project. In particular, we got the AI to provide multifaceted aid. From counting the user's number of full-motion reps to acknowledging and guiding proper form.
What we learned - Our team learned about Front end design, UI/UX, back-end, AI, and apart from all this, collaboration and leadership.
What's next for HB Fitness App - Deploy to the web, implement YoloV4, develop phone application, use Firebase for the database for users to create an account.
Built With
- figma
- flask
- google-drive
- jira
- mediapipe
- notion
- opencv
- python-anywhere
- tensorflow
Log in or sign up for Devpost to join the conversation.