This is a python code that detects the pulse of a an individual using a webcam or Network IP camera (http/ rtsp/ rtmp). This is a non-touch based method specifically for those people suffering from Hypersensitivity. Other than this, this software is extremely useful in current COVID-19 pandemic as it requires pure contactless functioning. This software is build using OpenCV Framework and using Fourier transformation to get the Pulese Value on a fitered and cleaned data received.
- This software uses OpenCV to detect human face and isolates the cheeks region.
- Cheeks region contain most of the blood vessels passing through and thus is in our Region of Interest(ROI).
- It monitors green color intensity(subimage of red colou, as red is reflected) in the ROI and isolates the pulse pattern from the average intensity.
- using a low bandpass filter and Fast Fourier Transformation.
- The exagerated color patters makes it easier for heartbeat identification.
- Then the pulse is isolated in about 5 secs and updates every 2 secs and syncs with user's heartbeat.
- execute
$ python run.py --camera 0 # webcam(number)
--source file.jpg # image
file.mp4 # video
rtsp://170.93.143.139/rtplive/470011e600ef003a004ee33696235daa # rtsp stream
rtmp://192.168.1.105/live/test # rtmp stream
http://112.50.243.8/PLTV/88888888/224/3221225900/1.m3u8 # http stream
--move True # If camera/user are not stable
False # if camera/user are stable- Wait for 5 secs for software to isolate Pulse. The waiting screen looks like this:
- The isolated Pulse value is shown in the screen and gets updated every 2 secs.
Face detection mode to be downloaded and kept in same repository, file to be downloaded : "shape_predictor_68_face_landmarks.dat"
- Add multi-Processing to enable faster detection and easier processing.
- Add a GUI support.
- Add multi-streaming/multiple face detection.
Made with ❤️ by Rohan Dubey