yonghaoxu/crgnet

[IEEE TIP 2022] Consistency-Regularized Region-Growing Network for Semantic Segmentation of Urban Scenes with Point-Level Annotations

crgnet's Issues

RuntimeWarning: invalid value encountered in divide

I encountered the following error, will this affect the training? I got very bad results.
/mnt/4t_disk/sxq/CRGNet4/utils/tools.py:98: RuntimeWarning: invalid value encountered in divide
acc_cls = np.diag(hist) / hist.sum(axis=1)
/mnt/4t_disk/sxq/CRGNet4/utils/tools.py: 100 : RuntimeWarning: invalid value encountered in divide
iu = np.diag(hist) / (hist.sum(axis=1) + hist.sum(axis=0) - np.diag(hist))
VaihingenSeg_log.txt

Where did you download the vgg16 pretrained weight?

Hi, @YonghaoXu. Your work is great and inspires me lot.
I want to know the link that you download the vgg16 pretrained weight.
I notice that pytorch has already provided a vgg16 pretrained weight in models.vgg16(pretrained=True).
So I am curious about why you use another pretrained weight.
Thank you very much.

About how to preprocess Vaihingen dataset labels！

Since the url of Vaihingen dataset has expired, and the labels of the dataset found on the Internet do not satisfy the current project, I would like to share with you how I deal with the labels.

Here is my code. Welcome star and follow me!

At present, I only uploaded how to preprocess labels and other places that need to be modified in the project, and I will update it later.