A python based home security observer based on OpenCV optimized to run on raspberry pi zero.
This program observes the camera video input and takes snapshots whenever it detects a person using OpenCV's body detection algorithm.
Here is an example of how the snapshot would look like:
$ python -m src.capture.security_cam -h
usage: security_cam.py [-h] [-d dropbox_uploader] [-v] [-w] [-l logdir] [-c crop]
[-p period]
snapshot_dir
Captures images and videos from camera based on person detection
positional arguments:
snapshot_dir directory for saving captured images
options:
-h, --help Show this help message and exit
-d, --dropbox_uploader Path to dropbox uploader script
-v, --verbose Set the log level to DEBUG
-w, --window Show video window
-l, --logdir Path to the log directory
-c, --crop A tuple of four integers for cropping frame
-p, --period Period in seconds for analysing video frame
If you wish to upload the images to dropbox, there is a great tool to upload there.
https://github.com/andreafabrizi/Dropbox-Uploader
You can sepecify the path to the dropbox uploader script with the optional --dropbox_uploader parameter.
Prequisite is ofcourse you have setup the dropbox uploader script properly and the device is connected to internet.
This tool has been be tested with and optimized to run on raspberry pi zero with a camera attached.
This tool is tested on python 3.9, 3.10 and 3.11.
Install the python dependencies with:
python -m pip install -r requirements.txt --prefer-binary
- Installing python dependencies on the raspberry pi can take considerable amount of time.
- Especially on raspberry pi zero, there can be specific issues installing opencv or numpy. This usually boils down to missing OS packages that need to be installed.
Please be aware of your local legal guidelines for installing security camera and taking people's images.