|
def segment(net, device, inference_loader, qlty_object, tiled_client): |
I think it is better to split this function into separate chunks:
- one overall function that get a single frame from tiled
- ship the frame (or frames) to the segment function. in the segment function we run qlty internally, segment stuff and return back in image in the right shape.
- push results back via tiled
I will write step 2 right now