Training on pre-existing images

Is there an equivalent of the camera-capture tool in jetson-inference, but for using a folder of images that have already been collected, instead of taking in a video stream? I’m trying to figure out how much of this I’d have to modify to get things working for this purpose, vs building something from scratch.


Jetson-inference do have some example that takes image as input.

You can modify the function here that allows the app to run for all the image within a folder: