Ever wanted to watch someone breathe using two cameras and a GPU?
No? Well, we did it anyway.
This repo uses stereo vision + disparity estimation (via AdCensus algorithm) + OpenCL acceleration to detect subtle chest movements — i.e., human breathing — in real-time.
Because... why not?
- 🎥 Two cameras = stereo vision
- 📏 Disparity map = depth perception
- 💨 Depth changes = chest movement = breathing
- ⚡ OpenCL = GPU speed boost
- 🧮 Based on the AdCensus algorithm (the census you actually care about)
- Real-time breath detection via disparity changes
- GPU-accelerated using OpenCL (no CUDA here)
- Stereo image capture and processing
- Works with low-cost cameras and mid-tier GPUs