A blend of Rehab + Habit → Empowering consistent recovery through AI.

Rehabit is an AI-powered rehabilitation assistant built to help stroke recovery and mobility-challenged patients perform guided physical therapy exercises at home.

By combining pose estimation, AI feedback, and voice coaching, Rehabit turns rehabilitation into a consistent daily habit — focusing on progress, balance, and confidence.

Stroke rehabilitation requires repetitive, supervised exercises — but most patients lack daily guidance or immediate feedback when practicing at home.
Rehabit solves this by providing:

• AI-driven real-time posture correction
• Voice and text feedback for form and symmetry
• Progress tracking and motivation through scores

• AI-powered pose tracking using MediaPipe to detect 33 body landmarks
• Real-time feedback analyzing body symmetry, range of motion, and form
• Voice and text guidance using Gemini and ElevenLabs
• Exercise modules including Arm Raises, Elbow Flexion, Sit-to-Stand, and Marching in Place
• Live overlays that show posture and movement cues
• Session reports to track improvement and accuracy over time
• Scoring system for performance evaluation

1. Frontend

   • Captures webcam input
   • Displays pose overlay and exercise guide
   • Streams video frames to backend in real time

2. Backend (Flask + SocketIO)

   • Processes frames using MediaPipe and OpenCV
   • Computes joint angles, range of motion, and symmetry values
   • Passes data to AI agent (Gemini) for analysis

3. AI Agents

   • Gemini interprets form accuracy and provides exercise advice
   • ElevenLabs converts AI feedback to natural voice for user motivation

Each exercise includes posture detection, form validation, and AI-generated audio coaching.

# 1) Clone
git clone https://github.com/KR13H/Rehabit.git

# 2) Virtual env + pip
python3.11 -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip

# 3) Install deps
pip install -r requirements.txt


# 5) Run
python app.py