This tutorial will show you how to setup and configure your own Vision AI solution using DeepStream, Telegraf and InfluxDB.
The goal:
- Jetson Device
- Web Cam
- Complete part 1 of the tutorial
- DeepStream 5.1 installed
- InfluxDB OSS installed on your Jetson (Ubuntu - ARM64)
- Mosquitto MQTT Broker (See Setup)
Please follow my blog here
Note: This setup guide will provided you with the needed steps to run the tutorial. Please check out the blog located under Background for a deeper understanding of the architecture.
- Clone the the project repository to your Jetson device:
https://github.com/InfluxCommunity/jetson_series_vision_pipeline
- Install the pip requriments for the project:
python3 -m pip install -r requirements.txt
- Update your apt repositry and then install the Mosquitto Broker:
sudo apt update && sudo apt install -y mosquitto
- (Optional) Enable the Mosquitto Broker to start on boot:
sudo systemctl enable mosquitto
- Start the Mosquitto Broker:
sudo systemctl start mosquitto
- Install InfluxDB OSS onto the Jetson and complete setup.
- Import both templates into InfluxDB OSS: Jetson_Stats & Inference_Template
- Update the following plugins:
- influxdb_v2 (edge_jetson_stats): Add token (OSS)
- influxdb_v2 (edge_inference): Add token (OSS)
- influxdb_v2 (cloud_jetson_stats): Add token (Cloud)
- Move the file to the telegraf directory:
sudo mv telegraf.conf /etc/telegraf/telegraf.conf
- Restart the telegraf service. Note make sure your InfluxDB Instance is ready:
sudo systemctl restart telegraf
python3 main.py /dev/video0
-
Create a Grafana account and Install the FlightSQL Library
-
Go ahead and upload the Vision AI Json file into your dashboard.
Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.
Please make sure to update tests as appropriate.