RepVal: A Real-time Exercise Form Correction System

RepVal is a system that uses computer vision to provide real-time feedback on exercise forms, leveraging MediaPipe to track body landmarks and a state machine-based approach to judge repetitions.

System Workflow

The overall workflow of the RepVal system is as follows:

A more detailed view of the system's logic for navigating through exercises and completing a routine:

System UI

Here is a showcase of the application's user interface.

Main Screen	Routine Builder	Live Camera Feed	Video Selection

Dataset: CFReps

This project uses the CFReps dataset, as we presented in our paper. The dataset can be downloaded from the following link:

(Download CFReps Dataset)

Setup

To run this project, you need to set up a Python environment with MediaPipe. Please follow the official (MediaPipe) installation instructions to prepare your environment.

How to Run

You can run the application in two modes: with a graphical user interface for live feedback, or in bulk processing mode for analyzing multiple videos simultaneously.

GUI Mode

To run the application with the live camera feed and user interface:

python mediapipe/main.py

Bulk Processing Mode

To process a directory of videos in bulk:

python mediapipe/bulk_process.py