Skip to content

blazinbanana/ACL-Risk-Predictor

Repository files navigation

ACL Injury Risk Assessment using Deep Learning

ACL Demo

Hugging Face Python TensorFlow OpenCV License Stars Issues


Live Demo

Try the ACL Risk Predictor on Hugging Face Spaces (You might need to restart the space)

Upload your lunge video and get AI-powered ACL injury risk analysis instantly.

  • Tips: Record your video in Landscape view. Ensure there is only one person per video for accurate analysis. You can use Front View and then Side View for a more comprehensive and accurate assessment. You can also source a video from YouTube!

Overview

Seamless ACL injury risk assessment from lunge movements using computer vision and deep learning.
Designed for athletes, rehab clinics, and biomechanics research.

Business & Market Potential

The ACL Injury Risk Predictor bridges the gap between sports science, injury prevention, and accessible AI-driven analytics. Designed for both athletic performance monitoring and clinical rehabilitation, the system provides actionable movement insights using only video input.

Target Users

Sports Medicine Clinics – For pre-season screening and return-to-play assessments.

Athletic Teams & Coaches – To monitor form and identify high-risk movement patterns early.

Rehabilitation Centers – To track progress post-surgery or during physiotherapy.

Fitness & Training Platforms – To integrate movement risk analysis for users and trainers.

Value Proposition

Prevent costly injuries – ACL tears can cost $5,000–$20,000 in surgery and recovery; early detection helps reduce risk.

Affordable biomechanics insights – Delivers motion analysis comparable to lab-grade systems at a fraction of the cost.

Scalable deployment – Works on ordinary cameras.

Data-driven decisions – Enable objective tracking of athlete performance and joint health trends over time.

Data

Front View and Side View videos – Recorded at Kenyatta University. We used two same level iPhones to maintain consistent Camera quality.

Target population – We had two groups: Athletic vs. Non-Athletic (Inclusive of both genders)

Initial Data Analysis – The data intended for training and validation were analyzed using the software Kinovea. Our key parameters were: Max Knee Valgus Angle, Max Knee Flexion Angle, and Max Trunk Lean Angle.

Labels – Using our analysis as well as consultation with the physiotherapy department, we now labelled our training data as High Risk and Low Risk

Sample Output

Demo

Video ACL injury risk assessment from lunge movements using computer vision and deep learning.


Features

  • Recorded Video Analysis: Process videos for instant ACL risk assessment
  • Biomechanical Feature Extraction: Automatic detection of knee valgus, flexion angles, and trunk lean
  • Deep Learning Model: LSTM-based classifier trained on movement sequences
  • Web Application: Streamlit interface for easy use
  • Reports: Gives a Risk Analysis to guide the user on next steps

Prerequisites

  • Python 3.9+
  • TensorFlow 2.8+
  • OpenCV 4.5+

Installation

git clone https://github.com/blazinbanana/ACL-Risk-Predictor.git
cd ACL-Risk-Predictor
pip install -r requirements.txt

About

Biomechanical Analysis System for ACL Injury Risk Assessment

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages