Hi.
The website said " # <object_id> is the same as the pixel values of object in annotated segmentation PNG files. "
However, when I load the PNG files, they have no <object_id>-like pixel value (e.g., '1' or '2' or '3').
Only RGB vector is exsisted...
How can i get right <object_id> from color annotation PNG files?
Thanks.
Posted by: YoungeunKim @ Aug. 7, 2018, 8:45 a.m.Hi, please try to read the images with index format. In Python, you can use PIL.Image.open(), in Matlab you can use imread(). Thanks.
Posted by: Chestnut @ Aug. 7, 2018, 4:27 p.m.Thanks! it works.
One more question,
I want the results have same colormap with examples, but when I save numpy array using PIL.save the results have grayscale.
For example,
----------code----------------------
from PIL import Image
result = Image.fromarray(segmentation_result.astype('uint8'))
result.save('00000.png')
--------------------------------------
Thanks a lot.
You can use something like this to save the image into index format:
https://github.com/linjieyangsc/video_seg/blob/master/davis_eval.py#L63-L64
Thanks.
Posted by: Chestnut @ Aug. 8, 2018, 1:52 a.m.Thanks very much for your help!
Posted by: YoungeunKim @ Aug. 8, 2018, 2:07 a.m.