🏭 Automotive Industrial AI Suite
This repository contains a specialized suite of Streamlit applications designed for Lean Manufacturing Motion Studies.
🚀 Applications Included
-
⏱️ AI Yamazumi & Motion Analyzer A high-performance Computer Vision tool that uses MediaPipe Pose Estimation to automate work measurement.
Live Motion Classification: Automatically distinguishes between Value-Added (Process) and Non-Value-Added (Walking/Waiting) actions.
WebRTC Integration: Optimized for use on Mobile (Android/iOS) and Cloud Servers using real-time browser-based streaming.
Takt Time Monitoring: Live visual alerts and "Target Takt" reference lines to identify line bottlenecks.
PDF Reporting: Generates a professional motion study audit report including cycle time history and performance status.
🛠️ Installation & Setup
- Requirements Create a requirements.txt file and paste the following:
streamlit
streamlit-webrtc
opencv-python-headless
mediapipe==0.10.11
pandas
plotly
reportlab
numpy
- System Dependencies For Streamlit Cloud deployment, you MUST include a packages.txt file to prevent OpenCV errors:
libgl1
libglib2.0-0
- Local Deployment Run the following commands in your terminal:
pip install -r requirements.txt streamlit run yamazumi_ai.py
📱 Mobile Usage Instructions Deploy the app to Streamlit Cloud.
Open the URL on your smartphone.
Allow Camera Access when prompted by the browser.
Press Start to begin the live AI Motion Study.
To finish a cycle, wave your Right Hand in the yellow "FINISH" box on the screen.
📊 Logic & Methodology Value Added (VA): Triggered when hands are raised relative to head position (simulating assembly work).
Walking (NVA): Detected via horizontal displacement of the torso/nose.
Waiting (NVA): Logged when no significant movement is detected for > 2 seconds.
Citation :
Mohd Rashidi Asari. (2026).
RashidiA/Yamazumi-AI-Analyzer:
Initial public release : Yamazumi Live AI Analysis (v1.0.1).
Zenodo. https://doi.org/10.5281/zenodo.18850820