Getting Started with the BrainFrame Devkit

First Steps

Getting a License

The first thing you will see on your devkit will be a prompt to upload a license.

To get a free no-expiration license, follow the instructions here. The devkit will need to have internet in order to validate the license. After following the instructions, you should see a client that is preloaded with a demonstration video.

Preloaded Demo

The BrainFrame installation on the devkit is a preloaded demo that has a video stream, several preloaded VisionCapsules, zones, an alarm, and a realtime dashboard visualization. After uploading a license, the client will open up and look like this:

If you click the video stream, you can get to a maximized view with more options:

This will provide an expanded view of the video, allowing you to better inspect the stream, and to change configuration relating to the stream.

Now, if you open Firefox, on the second tab there will be a dashboard showing realtime analytics for the demo video. This can also be accessed by going to the url http://localhost. The username/password for the dashboard is admin/devkit.

Updating the Devkit

Before going any further, let's make sure the devkit is up to date!

Info

The default password for the devkit is devkit

Update system packages

sudo apt update && sudo apt upgrade

Update the BrainFrame CLI

sudo -H pip3 install --upgrade brainframe-cli

Update the BrainFrame Server

sudo brainframe update

Update the BrainFrame Client

To update the interface, download the latest Ubuntu client. Then, install it by running the following, after replacing YOUR_DOWNLOADED_ZIP_HERE with the path to the downloaded client zip:

unzip -o -q YOUR_DOWNLOADED_ZIP_HERE -d ~/brainframe/brainframe_client/

Configure your own Stream

Setting up your first camera

Try adding a webcam! First, plug in the webcam and restart the server to make sure that Docker can see the camera. To restart the server, run:

brainframe compose down && brainframe compose up -d

To add the camera, follow the instructions here, and select the webcam option on the dropdown. To determine the ID of the webcam, you will need to run ls /dev/ | grep video. You should see video0 or something similar. If the result is video0, then put 0 into the client webcam textbox. The ID will probably be 0 if there is only one webcam connected.

Changing the Capsule Configuration

By default the devkit ships with the capsules necessary to run the preloaded demo:

These capsules can be found in the capsules directory, which is in /var/local/brainframe/capsules. More information about loading new capsules can be found in our 'Using Capsules' tutorial.

General Usage

Controlling the Server

More information about general usage of the CLI can be found here.

To stop the server

brainframe compose down

To start the server

brainframe compose up -d

To restart the server

brainframe compose down && brainframe compose up -d

Advanced

Server Configuration

For all of the possible server configuration variables, refer to here. On the devkit, the .env file is in /usr/local/share/brainframe/.env

cat $(brainframe info install_path)/.env

AI accelerator configuration allows you to customize which AI accelerator your capsules use. NVIDIA GPU is enabled if it is available. AI accelerators can be enabled on Intel chipset with OPENVINO_DEVICE_PRIORITY environment variable setting. By default, it is set to HDDL,CPU, meaning that all capsules will load onto the HDDL and CPU devices, giving priority to HDDL. Since the devkit has 2 MyriadX devices, this means that having more than 2 OpenVINO capsules will make the inference run slowly. If you want to experiment loading many capsules, set OPENVINO_DEVICE_PRIORITY=CPU.