sssAI

AI based motion detection for Synology Surveillance Station - For instructions on use see https://blog.cadams.me - Wiki coming soon!

Features:

• Uses https://deepstack.cc/ for object recognition
• Use the native DS Cam app for mobile notifications
• HomeKit integration (via HomeBridge and Homebridge Webhooks & Camera-ffmpeg plugins)
• Captured image with border-box annotations saved for review

Builds upon work by Christopher Adams (Christofo - original design), CoooWeee (CoooWeee - Ignore Areas feature), and Thiago Figueiró (thiagofigueiro - Docker Compose support) New Features I will be looking into:

• detect_labels moved out of settings.json. detect_objects replaces detect_labels and is now per camera.
• Adding feature "ignore_polygons". This builds upon ignore area which lets you ignore an rectangular section of your camera image; additionally we will be able to ignore a polygon. "ignore_polygons" will be per camera per detect label
• Also changes to allow min size for each type of detect type. Each detect type (e.g person, car, etc.) will have its own set of polygons to ignore and these values can be different per cam. Same goes for ignore_areas.
• Windows or web app tool to be built in future to allow user to grab snapshot from SSS draw ignore polygons and export JSON for your config (TBD).
• Push notifications by calling docker push container directly from Python
• Storing video clip in cloud (likely google) which can be viewed directly from push notification

Performance

• DS920+ (20GB RAM) - Deepstack set to "low" - ~2 seconds for image recognition
• DS918+ (12GB RAM) - Deepstack set to "medium" - ~4 seconds for image recognition
• DS713+ (4GB RAM) - Deepstack set to "low" - ~9 seconds for image recognition

Running using docker-compose

If you want to run the containers in a docker host separate from your Synology device, you can use the docker-compose.yml file.

1. Clone this repository on your docker host
2. Create the required directories (mkdir -p ./data/captures ./data/deepstack)
3. Create settings.json and cameras.json (see *.example files)
4. Optionally edit docker-compose.yml to configure your time zone and network ports.
5. Start the services (docker-compose up -d)
6. Optionally tail the log files (docker-compose logs -tf sssai)

On the first time you run the service