Client-server model to detect human fall detection using OpenPose computer vision library
This is a simple project that I created when I explored computer vision using Open CV. It uses OpenPose for detecting human like figure in your video feed.
This is a client-server model ( client records and sends []using TCP] video to server for processing). I tested it over WLAN. Explore futher if your requirements demand more.
-
You need python installed.
-
You will need to install tensorflow.
-
You cant install cv2 library. Instead install "opencv-python". Other dependancies should be easy to figure out.
-
OpenPose is the library used. Refer to the youtube series below for setup instructions. ( https://www.youtube.com/watch?v=4FZrE3cmTPA 3 videos from here) -- This is the 😅 toughest part of project. Setting up OpenPose.
-
Place the project files in the folder [ pose\tf-pose-estimation ]
-
You can alter the fall detection logic as you needed by tweaking the code.
-
The cam that needs to be used can be configured in OpenCV function.
-
Check firewall settings for server if you use the project in the second way.
-
client file should be run from command prompt. It might not by default work if u run it from pycharm or any editor.
Now the ball is in your court.