milleniums / High-Resolution-Remote-Sensing-Semantic-Segmentation-PyTorch

PyTorch实现高分遥感语义分割(地物分类)

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

error: no module name gid_classes

AirHighway opened this issue · comments

from class_names import gid_classes

In the dataset.py file, the line 4:

from class_names import gid_classes

But the file class_names.py just have the def fifteen_classes and def five_classes functions, where are the gid_classes?

commented

Hello, has the problem been solved

The GID is a dataset. PAPER DATASET
More information you can capture in the README.md.

I think the gid_classes is a function that you choose by yourself, how many classes land cover you want to show, then you will choose what function. Of course, it maybe determined by the data label.

The function def fifteen_classes and def five_classes mean that the label has fifteen classes land cover and five classes land cover, separately. They are examples of the gid_classes.

Here, I review the train.py and know the classes is default five. So you try change the from class_names import gid_classes to from class_names import five_classes. Also, you should correct the codes in line23 in dataset.py which self.class_names = gid_classes() to self.class_names = five_classes(). If through this step, the codes also can not run, I suggested that you read more detail about the codes. If you find the way solve this problem, I'm glad to communication with you.

@ROY-SNN I have ask the question, but some experiments should you do self.