For beginners starting their fitness journey, staying consistent and knowing what workouts to do or how to do them properly is one of the biggest hurdles. That's where XRSZE (pronounced "exercise") comes in.
XRSZE addresses this problem by using a computer vision algorithm that leverages posture detection to accurately count reps and verify proper form. We also used the Groq LLM API to build an AI chatbot trained to tailor workout and nutrition plans based on user-specified fitness goals and parameters such as height, weight, age, and dietary restrictions.
We built XRSZE using various web technologies and APIs. The front-end interface was designed with HTML/CSS/JS. We used Node.js for the backend and integrated it with MediaPipe and TensorFlow for computer vision and movement tracking. We also specifically calibrated the MediaPipe recognition model to detect accurate body movements that will recognize workouts and repetitions.
One of the biggest challenges was ensuring the accuracy of movement tracking with MediaPipe and TensorFlow. Fine-tuning the model to reliably count and recognize various exercises required extensive testing and adjustments. We also faced difficulties integrating the Groq API seamlessly with our backend, leading to several iterations before proper communication between the chatbot and the user. Optimizing the front-end performance to handle real-time data without lag was crucial for a smooth user experience.
We learned a lot about integrating complex technologies, like MediaPipe, TensorFlow, Node.js, and Groq to work together in a unified system. The process taught us the importance of iterative testing and refinement, especially when fine-tuning AI models for specific applications. We also gained valuable insights into optimizing real-time data processing and enhancing user experience through thoughtful design and responsive feedback mechanisms.
We’re excited to expand XRSZE into a mobile app, and we are already working on the Flutter code to make our platform more accessible and convenient for users on the go. We also plan to enhance our workout tracking capabilities by adding support for a wider variety of exercises and workouts, ensuring that users have a comprehensive tool to meet their fitness goals.