Hi! I’m a newbie using Jetson Nano. I want to ask, if any suggestion or optional to send data from Jetson Nano to the IoT server? For example, I have data for object detection using Jetson Nano but I try to figure out how to send this data to the server and want to convert it to .json format. Is it possible? Thank you.
You could check out Deepstream 6.0. You can provide it with an ONNX formatted network network for object detection and it can stream the results back to your server over UDP. TensorRT will take care of optimizing your network for inference. And there are a ton of community built tools you can use to make it run even smaller and faster.
You can even add secondary classifiers to the pipeline. ie: it will cut out a bounding box from an object detector, then run it through further classifiers.
NVIDIA provides code samples for object detection pipelines.
Prepare to do a bit of work though. There is a lot to get your head around, but I think it’s worth it myself.
1 Like
Also, Deepstream does seem to support converting object detection to json format with the Gst-nvmsgconv plugin.
Gst-nvmsgconv — DeepStream 6.1.1 Release documentation and sending it to a variety of different servers using the Gst-nvmsgbroker plugin Gst-nvmsgbroker — DeepStream 6.1.1 Release documentation
1 Like