SAVCAM (Smart AI Viam Camera) Flutter App

The SAVCAM app uses Viam's Flutter SDK to control settings for a highly configurable security camera leveraging the Viam Computer Vision service, allowing for configurable rules that use any configured detectors or classifiers (for example, machine learning).

This app provides a way to configure event and notification rules for the SAVCAM event manager, and provides a front-end to view live camera streams and camera replays of triggered events.

Setup

Hardware and Viam configuration

For hardware, any number of cameras (at least one) compatible with Viam are required. We have tested with an RTSP camera as well as a USB webcam, but any should work. You will also need a computer running 64 bit linux or MacOS. We have tested with a Mac laptop and a Raspberry Pi.

Go to the Viam App and configure a new machine. Follow setup instructions, then go to the Config tab in the Viam app. Set up a camera component of the model you will be using. Our RTSP camera was set up with model 'ffmpeg' with this configuration (note, we use a fairly low resolution to reduce processing requirements):

{
  "filters": [
    {
      "name": "scale",
      "args": [
        "640:480"
      ]
    }
  ],
  "video_path": "rtsp://10.0.0.29/1"
}

Next, in the Services section of the Config tab, you'll likely want to configure one or more Vision Services. This could be an ML Model-based classifier or detector, or a pre-packaged vision service from the Viam registry like a facial detector or motion detector.

Once configured, these Vision Services will be available for use by the SAVCAM event manager, which you will not configure from the Components tab (choose event-manager from the 'Create component' menu). You can leave the configuration attributes empty, but choose any Vision Services you may want to use as resources your event-manager "Depends on". Save this configuration.

Finally, you'll need a replay camera of type 'image-dir', so select and configure that from the 'Create component menu', and save. This will allow you to play image loops of any triggered security events that you configure.

API Keys and mobile app configuration

This example app requires the use of API Keys. You can obtain an API Key by following the documentation. You will also need your configured robot location, part id, and the name of the event-manager and image-dir camera you configured.

Hardcode your API Key and other robot information into a .env file, should look like:

API_KEY_ID=<API KEY ID>
API_KEY=<API KEY>
ROBOT_LOCATION=<ROBOT LOCATION>
EVENT_MANAGER=<EVENT MANAGER COMPONENT NAME>
DIR_CAM=<IMAGE-DIR COMPONENT NAME>
PART_ID=<PART ID>

Permissions

On many operating systems, apps are not granted access to the internet by default. To update these permissions, follow the steps below for whichever operating system you are targeting.

Android

Android apps must declare their use of the internet in the Android manifest (android/app/src/main/AndroidManifest.xml):

<manifest xmlns:android...>
 ...
 <uses-permission android:name="android.permission.INTERNET" />
 <application ...
</manifest>

macOS

macOS apps must allow network access in the relevant .entitlements files (macos/Runner/{DebugProfile | Release}.entitlements).

<key>com.apple.security.network.client</key>
<true/>

Run

To run the app, you should run the lib/main.dart file:

flutter run -d macos lib/main.dart

Screenshots

• Home Screen: Home screen, where you will see a preview of any configured cameras and any triggered alerts.

• Camera Stream Screen: Screen where you can see a live stream from any configured camera.

  • Event Configuration Screen: Screen where you can see and manage configured events.

  • Home Screen with Triggered Alert: Home screen shows the most recent triggered alerts and their video preview.