The ADLINK Edge® Deep Learning Streamer processes input data, performs inferencing on the data and publishes results on the ADLINK Data River for further processing. This guide explains how to configure and deploy Deep Learning Streamer to the Vizi-AI and how to display the inference results using a demo stream in real-time.
The app is set up to obtain a sample video from Github and inference models are included to detect age, gender, emotion, identifying features and head position.
As well as Deep Learning Streamer the profile also includes the following applications:
- Node-RED – To provide publish and subscribe capability to access ADLINK Data River in Node-RED.
- Portainer – To manage the container via portainer, this is useful for obtaining log files.
Ensure the Vizi-AI is set up prior to following these steps. Refer to one of the following:
Note: The inferencing results are output via RTSP, therefore do not show within the Vision tab in Edge Profile Builder. In this example we use VLC Media Player to view the inferencing results.
Download the Deep Learning Streamer Profile
- Start your web browser, open Profile Builder from http://localhost:8082 and click the Projects tab.
- Click Create project and choose a “Project Title” and “Description” for your profile and click Create.
- Within Projects, click Download Profile, choose ‘ADLINK-Templates’ from the drop-down in the top left of the window, locate and select the ‘intel-deep-learning-streamer-x86′ profile and click Next. Choose the project and optional group to download the profile to and click Download. Once the download is complete, click Finish.
The profile shows as follows:
Deploy the Profile to the Vizi-AI
Within the project, select the profile and click Deploy, if your device is registered locally, select Deploy directly to a device, click Next choose the device and click Deploy.
If your device is registered with Azure IoT Edge, select to Upload to a cloud service, click Next, choose the service to upload to and click Upload.
View the Inference Results
To view the inference results, use a third party video streaming application, for example, VLC Media Player. The RTSP address is “rtsp://<DEVICE IP ADDRESS>:8554/intelDemoStream” where the <DEVICE IP ADDRESS> is the IP address of your Vizi-AI, for example: “rtsp://192.168.0.66:8554/intelDemoStream”.
In VLC Media Player select Media and Open Network Stream, enter the RTSP address and click Play.
This shows a demo stream of a man and a woman with inferencing for their age, gender, emotion, identifying features and head position.
If the video does not load it may be that the video has ended as it is not on a loop. You must restart the Deep Learning Streamer application to restart the video:
- In Edge Profile Builder select the Devices tab and click the Portainer icon on the Vizi-AI.
- If you have already created a Portainer login, use these credentials to log in. If not, enter the username details, click Create user, select Docker and click Connect.
- Within Portainer click Containers, select the check box for deep-learning-streamer and click Restart. After a few seconds the video stream is available again.