-
-
Pose detection
-
Title screen
Inspiration
Every day, millions die from obesity-related health issues, with over 2.8 million deaths annually linked to obesity worldwide. Hospitals are overwhelmed, and nearly 50% of gym-goers suffer injuries due to improper form or lack of guidance. The wealth disparity also prevents many from affording personal trainers, leaving them without the support they need. That’s where LiftMate comes in—your personal gym buddy, powered by AI. It provides real-time form feedback, personalized workout plans, and diet advice, ensuring you train smarter, prevent injuries, and stay on track. LiftMate is here to make fitness accessible, safe, and effective for everyone, no matter your budget
What it does
Liftmate is a comprehensive fitness companion that ensures you perform exercises with proper form, minimizing the risk of injury. It provides real-time feedback and personalized tips to help you refine your technique and optimize your workouts. Additionally, Liftmate features an intuitive chatbot that offers expert advice on creating customized workout plans and balanced diets. Whether you're seeking guidance on fitness goals or simply need someone to chat with, Liftmate is here to support you on your fitness journey.
How we built it
We built Liftmate by leveraging the power of the YOLOv8 pose detection and custom trained classification models to accurately assess and analyze users' form during workouts. Our frontend, built with React and Vite, provides a smooth and intuitive user experience, ensuring the app is both welcoming and easy to navigate. To evaluate form, we calculate the angles between joint points detected from the user's movements, offering precise feedback on posture and technique. The backend is powered by Flask, ensuring seamless communication between the frontend and the AI models to deliver real-time feedback and advice.
Challenges we ran into
The challenges we encountered during the development of Liftmate were primarily related to integration and time constraints. We had to deal with very little compute while training our custom classification model. One major hurdle was integrating our classification model into the Flask backend, and we struggled to debug the issue despite extensive efforts. Additionally, due to time limitations, we weren’t able to implement the full scope of our project, leaving several exercises incomplete. Our form-checking feature also didn’t achieve the level of accuracy we were aiming for, and we faced performance bottlenecks with the pose detection, which we wanted to optimize for faster results. Despite these challenges, we learned a lot and are excited to continue improving the app.
Accomplishments that we're proud of
We’re definitely proud of the sleek and user-friendly front end we were able to create, which provides a smooth and intuitive experience for our users. Despite our limited experience with Flask, we’re also proud of building a solid backend that managed to support the functionality of the app, even if it wasn’t always flawless. Another key accomplishment was deploying our models, something we had never done before, and it was exciting to see our work come to life in a real-world setting. These milestones were significant learning experiences and gave us a sense of achievement.
What we learned
We learned a lot throughout this project, especially about project management and the importance of working in consistent environments. We encountered several dependency issues that could have been easily avoided with better communication and coordination. This experience highlighted the importance of keeping everyone aligned and ensuring we’re all using the same setup. Additionally, we gained more experience with Git, improving our version control practices and collaboration within the team. These lessons will definitely help us in future projects.
What's next for LiftMate
Next for Liftmate, we’re focused on expanding the range of exercises to provide even more value to users. We also plan to optimize the app's performance, particularly improving the speed of pose detection and enhancing the accuracy of form checks. These updates will ensure Liftmate continues to evolve and offer a more seamless, effective experience for users.
Log in or sign up for Devpost to join the conversation.