Evaluating on my own data about fine-grained-segmentation-networks

[maunzzz]

Hi!
Probably not, I believe the images are saved without any colormap so they might look just black in an image viewer.
You can try loading them and look at the values.

[kxhit]

Hi! @maunzzz Thanks for your reply.
I print the value of saved images and got all zeros.
Should I notice something when using segmentor.run_and_save()? Could you give me an example .py of inferencing one image?
I haven't got a clue what to do. Thanks a lot!

[maunzzz]

Hi, sorry for the late reply. Hopefully you solved your issue by now.
If you haven't, I would suggest looking at the Segmentor class in utils/segmentor.py
The run_and_save accepts a path to an image so it should be straightforward to do inference on one image.