Ropey-Cam is a Python-based surveillance camera system, designed for Raspberry Pi single-board computers with attached camera modules. The system provides real-time, low-latency, MJPEG video streaming to web browsers, while simultaneously monitoring for motion, and recording high-quality H.264/.mp4 video clips when motion is detected.
As a long term user of the comprehensive RPi_Cam_Web_Interface with various RPi cameras (V1,V2 and HQ), it was frustrating not being able to use third party cameras or any of the later RPi camera modules with RCWI. Unable to find any libcamera/Picamera2 based alternatives, and as a learning exercise, work started on a basic -very basic- alternative.
For those looking for more comprehensive features, there is available the very fully featured raspi-cam-srv.
"Ropey-Cam" reflects its construction methodology: it is assembled from code snippets and examples from the Picamera2 repository, with minimal Python complexity, and uses a multi-threaded architecture to handle concurrent operations. In it's basic form, when installed on a RPi with the standard desktop OS, the only additional dependency is OpenCV.
On a Raspberry Pi running a desktop RPiOS and with camera attached, open a terminal and enter :-
sudo apt install python3-opencv -y
git clone https://github.com/sandyol55/Ropey-Cam
cd Ropey-Cam
./Ropey-Cam.py
Then from another computer on the same network browse to raspberry.pi.ip:8000, typically an address like 192.168.0.120:8000.
(Or, on the Raspberry Pi computer, open a local browser and request 127.0.0.1:8000 - or even do both - to get simultaneous streams to multiple browsers.)
The browser should present a home page as below
The figures in the top left of the streaming video frame indicate the measure of (mse) frame to frame change in the video, against the currently set trigger level. If motion (or noise) in the video is present, in excess of the trigger level, then video recording will be activated as indicated by the red REC stamp in the top right.
If recording is activated when no apparent motion is present in the video, pressing the Inc_TriggerLevel button repeatedly will increase the frame to frame change threshold at which recording is triggered.
Similarly if motion in the frame does not trigger recording press the Dec_TriggerLevel button repeatedly until recording is triggered.
Once calibrated for the motion/noise environment, test that deliberate motion triggers recording.
The REC stamp will remain for ~ 6 seconds after motion has dropped below the trigger level, while the full video file - containing ~3 seconds pre-trigger, X seconds of motion and ~3 seconds of post-motion video - is saved to disk.
This recording is happening in the background while the live stream continues uninterrupted.
Brief description of the effect of the control buttons.
Motion_Detect_Off
will disable the motion detection, but retain the live stream.
will be replaced by Motion_Detect_On
Inc_ and Dec_TriggerLevel
as described above, modify the sensitivity to motion
Manual_Recording_Start
will initiate a recording irrespective of motion triggering.
will be replaced by Manual_Recording_Stop.
DELETE_ALL_FILES
will delete all the timestamped .mp4 video and associated trigger-point .jpg files that are stored in the Videos subdirectory.
Requires to be confirmed with a second press.
EXIT
will end the Ropey-Cam program on the Raspberry Pi,
also requires confirmation with a second press.
RESET
acts as a cancel button for any of the 4 other system buttons,
to allow the first press of those buttons to be ignored and reset.
REBOOT
will reboot the Raspberry Pi.
Best used in an advanced set-up where Ropey-Cam has been set to start on boot up.
SHUTDOWN
will shutdown the Raspberry Pi.
Like EXIT and REBOOT there is a built in delay to allow any in-progress recordings to be stored cleanly.
Typical use is expected to be in a headless remote mode.
Before deploying it will be useful to examine the code and, if necessary, make changes to match the Pi and camera being used, e.g.
To select the aspect ratio of the sensor being used.
To select the resolution of the main video recording stream and the browser stream
To select the required framerate
And to select the mode in which to operate the sensor.
These are controlled by documented commented lines within the code at ~lines 33-49, line 52 and line 80.
Edit as required.
The default configuration is currently :-
16:9 output with a 1280x720 main stream resolution and 512x384 for the browser stream, and a framerate of 25fps.
The sensor default mode is set to 1, which will typically give a full frame (16:9) 2x2 binned mode on HQ and V3 cameras, or a full frame (4:3) 2x2 binned mode on V1 or V2 cameras.
Other recommended steps before deploying are:-
set the Raspberry Pi up as a Samba server
set the Ropey-Cam program to start on boot, with a systemd service.
(Lots of online tutorials for these steps, but links to guides for creating both of these services are included at the bottom of this page).
As an example the screenshot below is taken from a Pi5, accessing Ropey-Cam running on a Pi3 Model B V1.2, with both machines connected via Wi-Fi to a home network.
In the upper left is an ssh session running htop on the remote Pi3 during an active recording, indicating useful spare CPU and memory capacity.
In the lower left the Thunar file manager is accessing the Videos subdirectory on the remote Pi3 via the network samba share. (Using a separate file manager from the inbuilt pcmanfm allows the thumbnail icon sizes to be increased without affecting the desktop).
And on the right is the browser, showing an image from the camera while it is being pointed at the browsing computer's screen - with a stopwatch in the foreground.
The difference between the stopwatch time in the foreground, and the background - earlier - image of the stopwatch that has been captured, processed, passed through the network to the browser and displayed, is a measure of the system latency. The latency of 170ms is in the mid-range of the values seen in this particular configuration, which were typically in the 150 to 190 ms range.
Much of the code is taken from the examples, discussions and responses to issues as found in the Picamera2 Github repository .
The method of button control was inspired by this article .
And signag offered coding suggestions against an early version of the code.