dahyun-kang/ifsl

its about dataset

ASSicat opened this issue · 1 comments

Hello, Thank you for your work.
If I want to use my own dataset for training, what should I do?

Hello,

We currently do not support the custom dataset for training.

It can be mannually done by replicating dataset/pascal.py.
First, a set of images and a set of mask-labeled annotation in *.png (pixel-level class id) are required.
And place them in separate folders so that they can be read like:

self.img_path = os.path.join(datapath, 'VOC2012/JPEGImages/')
self.ann_path = os.path.join(datapath, 'VOC2012/SegmentationClassAug/')

Best,
Dahyun.