LeanLens
Team Sawit Koseeyaumporn
The big news
Factories can finally see the time they're losing. LeanLens reads existing station cameras and recorded clips and classifies activity as value-added or waste, such as motion, waiting, searching and transport. It flags safety events and proposes Lean Six Sigma DMAIC fixes with estimated time saved per shift, which the operator approves. Instead of slow, expensive manual time studies, CCTV that was only used for security becomes a continuous improvement tool, helping manufacturers boost efficiency.
Watch the demo
This video can't play here. Use the "Watch the video" button below.
What it does
An industrial computer-vision tool that analyses CCTV footage with YOLO, BLIP and a Qwen vision-language model to classify work activity, detect safety events and suggest DMAIC improvements, with ElevenLabs voice output. Built with Python, FastAPI, PyTorch, Ultralytics, OpenCV and a Svelte and Vite front end, packaged with Docker.
Try it yourself
U-tags
Tap a tag to find similar projects
Meet the team
This is an independent hackathon project. Some features may be prototypes. Please do your own research before relying on it.