Jetson Nano Object Detection Networks - Object Classes

The object classes are well known for these Object Detection pre-trained networks: ssd-mobilenet-v1, ssd-mobilenet-v2, and ssd-inception-v2.

https://github.com/dusty-nv/jetson-inference/blob/master/data/networks/ssd_coco_labels.txt

But having a tough time finding the object classes for these networks:

coco-dog
coco-bottle
coco-chair
coco-airplane
pednet
multiped
facenet

Hi,

Most of these models are single-class detection model.

Ex.

Model: coco-dog
Class: dog

You can find the detail informaotn here:
https://github.com/dusty-nv/jetson-inference/blob/master/docs/detectnet-console-2.md#pre-trained-detection-models-available

Thanks.

Is it possible to use multiple networks in the same C/C++ program, such as this Nvidia detect program?

https://github.com/dusty-nv/jetson-inference/blob/master/examples/detectnet-camera/detectnet-camera.cpp

Hi xplanescientist, yes you can creatr multiple instances of detectNet object. You will want to use this overload of detectNet::Create() or similar, as opposed to the version that accepts argc, argv (command line).

Also you will want to call detectNet::Process() with OVERLAY_NONE and then perform the overlays (if desired) with detectNet::Overlay() after all instances of Process() for that frame. This is so that each detectNet instances Process() call is using the original frame, as opposed to one with bounding boxes/ect drawn on it.