Standalone mode for depthai_movenet

akhil

Hi, can this DepthAI MoveNet code be modified so that it can be flashed and run on standalone mode? I see this repo talks about an edge mode, but ideally we want all the code to be running on the camera and outputting its results through a TCP connection.

geaxgx/depthai_movenet

@RakshithSingh

lovro

Hi,
yes, this is possible with a PoE camera. What you're looking for is Standalone mode, where the app runs directly on the device at boot and sends results (e.g. via TCP) without needing a host.

Note: The “Edge mode” in depthai_movenet just means the inference runs on the device, but it still requires a host to launch the pipeline. Standalone mode is different. More info and setup guide: https://docs.luxonis.com/software/depthai/standalone

akhil

Hi, thank you for the reply. Does this mean the depthai_movenet lib can't be converted to a flashable format for standalone mode? I have this library executing in edge mode with a desktop host, but I was not sure how to convert it to a standalone application.

lovro

Hi,
It’s likely possible to run depthai_movenet in Standalone mode if you're using an RVC2-based PoE camera and the code is fully on-device.

Take a look at the official guide and example to see what would be needed:

Standalone mode

Example