Python >= 3.5
Parallel motion detection by simply running a motion detection tool over multiple videos at the same time.
The script will by default parallelize up to your machine's CPU cores and output separate compilation videos of all the motion detected events per video recursively in the supplied folder.
Download and install Miniconda or Anaconda.
Miniconda is a more lite version without the extra GUI applications.
Hit the Windows key and search for "Anaconda Prompt" and open it. A terminal prompt will appear.
Enter the following to create a new environment:
conda create --name dvr-motion
Enter the following to activate the environment (You must do this everytime you close the terminal):
conda activate dvr-motion
Install pip:
conda install pip
Install dvr-scan:
pip install dvr_scan[opencv]
For all systems. Simply install dvr-scan with pip:
pip install dvr_scan[opencv]
Check the help for the options:
python dvr_motion_parallel.py --help
Run with all defaults:
python dvr_motion_parallel.py path/to/videos/folder
You may want to specify a different output folder:
python dvr_motion_parallel.py path/to/videos/folder -o path/to/output/folder
Change the configs inside the dvr-scan.cfg file.
You can check all the options with dvr-scan -h, but the following should be
mainly the command that would be used:
dvr-scan -l 10 -i <path/to/video>
Also, try the following as supposedly this background subtractor may be faster:
dvr-scan -l 10 -b CNT -i <path/to/video>
You can tweak the threshold a little to try to filter out non-desirable detections:
dvr-scan -l 10 -b CNT -t 0.25 -i <path/to/video>
0.15 is the default threshold. A higher threshold would require more movement before a detection is triggered.
This will output several motion detected .avi clips in the current folder.
Best way is to manually choose 4 or more clip intervals in the long video that allows each clip to be more or less equally sized without splitting on a target object.
Create a times.txt file and put the split points
in this format:
HH:MM:SS[.nnn]
Look at times_example.txt for an example.