An example of capturing a video using a ring buffer
This is a basic demonstration of saving camera frames in a ring buffer, then saving them to a video clip when a machine fault happens. The result will be a video clip leading up to when a fault condition is detected. OpenCV is used to capture the video frames, pylogix is used for monitoring a tag for a fault condition. Webcams and IP cameras are supported.
The camera definition in the config is a dict, where the key is the camera ID (just used in the file name) and the path to the camera. Local cameras (built in web cams or USB cameras) will be an integer. In my case, I have a built in webcam and a USB camera, they are index 0 and 1 respectively. My particular model of IP camera is at the path: rtsp://ip_address/streaming/channels/1. See your manufactures documentation for your specific cameras path. I like to use VLC to verify that I have the right address for my IP cameras. In VLC, go to Media > Open Network Stream, paste the path to your IP camera and play it. If it successfully displays the camera, the path should work for recording the video.
When launched, an output directory will be created, in side of it, a directory for each camera that was defined in the config file. Video clips will be saved in the camera directories. You can define in the config file the maximum number of video clips you want to keep, the oldest will be purged with each new clip recorded.
CPU utilization is pretty low with this method since the buffer is saved in memory until the request is made to save the video clip. So the number of cameras, length of buffer and the resolution desired will determine the amount of memory that will be used while running. The default config values will use something like ~600MB of RAM. I have ran it on a Raspberry Pi 4, it performed reasonably well.
- python 3
- opencv-python
- pylogix
To run, make sure you setup your configuration file, then:
pylogix@pylogix-kde:~$ python monitor.py