Sense's edge processing software is called a node. Its main features include the ingestion of video data for training data creation and the deployment of an ML model over live video data.
The Sense edge node software requires processing power and should ideally be run on devices that meet the minimum requirements. Actual requirements will depend on your use case, the model(s) you are running, number of cameras involved, etc.
Minimum requirements to run a Sense node:
Linux operating system, preferably ubuntu 20.04
8 cores and 16GB of ram
Network of 1Gbit/s
latest version of docker installed and also NVIDIA/nvidia-docker
We support NVIDIA Jetson Xavier NX and AGX. Based on our testing, the Jetson Nano may be underpowered for many use cases.
Once your node is running, you can begin to add cameras to the node and start capturing video segments to create training data.