MoCam

Easy setup for motion detection camera compatible with next generation 64-bit machines.

MoCam is an implementation of motion-project.github.io for capturing video and images when motion is detected and Rclone to quickly upload captures to a cloud provider of your choice.

Getting started

1. Clone this repo onto the machine with camera connected

git clone https://github.com/xcollantes/mocam.git

2. Setup saving video and images to a cloud provider

Create an rclone.conf file to be able to upload captured media to a cloud provider of your choice. See list of 60+ cloud storage providers: https://rclone.org/#providers

Some providers include:

1. Install Rclone locally: sudo -v ; curl https://rclone.org/install.sh | sudo bash
2. Run rclone config and follow the prompt to generate an rclone.conf
3. Generated config can be found in ~/.config/rclone/rclone.conf. Copy or move file to mocam/ repo directory like so: mocam/rclone.conf.

(https://rclone.org/install)

3. Build Docker image

1. Review motion.conf. The default settings are calibrated for a USB camera on a RPI4.

   See motion-project.github.io for full options.

2. Build Docker image:

docker build -t mocam . 

4. Run Docker image

Make sure you have the following:

1. Connect camera to machine
2. Get your camera device name by running v4l2-ctl --list-devices on your machine.
3. Run command:

docker run --rm --device=MY_DEVICE_NAME --env RCLONE_DEST=MY_RCLONE_PROFILE:/MY_DIR MY_IMAGE_NAME

• --rm Remove the container if stopped.
• --device Docker flag to let container use the host machine's camera.
• MY_DEVICE_NAME
• --env Set ENV variables.
• RCLONE_DEST Rclone destination location with format as MY_RCLONE_PROFILE:/MY_DIR.
• MY_RCLONE_PROFILE Rclone profile name as founc in the rclone.conf file created.
• MY_DIR Folder in the cloud provider you setup. This can be a path with nested folders.
• MY_IMAGE_NAME Docker image name set in Docker build step.

Example:

docker run --rm --device=/dev/video0 --env RCLONE_DEST=remote-mega:/mo_cam mocam

1. See machine logs at /motion/log/motion.log if enabled in the motion.conf. You will also see the logs from the container by using docker logs CONTAINER_HASH

How it works

MoCam depends on Motion to capture video and images when motion is detected and Rclone to quickly upload captures to a cloud provider.

Common pitfalls

How do I get my device name?

Run v4l2-ctl --list-devices to see the devices attached to your machine.

Processor compatibility: arm64

Since this was tested on RPI4 which is uses arm64 processor, by default the Docker base image is python:bullseye which works with arm64 processor.

Previously, raspberrypi4-64-debian was used as the base image which worked with no issue.

Choose a base image which matches the processor of the machine.

motion.conf calibration

Depending on the camera's environment, the camera, the procesing power of the machine, and the network the camera is on, the Motion config might need to be adjusted. The settings and their values are most explained at https://motion-project.github.io/motion_config.html.

Below are the calibration notes I took for my setup:

My goals in order of priority:

1. capture accuracy (does the camera capture the subject fully?)
2. image quality (is the subject's face visible?)
3. network conservation (can the videos upload to the cloud in under 2 seconds?)

• framrate setting is MAX and seems to fluctuate
• I'm using event_gap of 1 for testing so upload occurs right away