attila94/ciconv

Original code for "Zero-Shot Day-Night Domain Adaptation with a Physics Prior"

ciconv's Issues

About Tokyo24/7 dataset

Excuse me, could you please tell me where to download the Tokyo 24/7 dataset? The download link I find on the original paper is invalid. So where you download it? Thanks

Regarding 3_segmentation pretrained models

Hi,

thanks for posting the code and models. I was trying out the 3_segmentation code folder. I was testing with refinenet and W-refinenet on night time driving dataset. I was getting 33.5 and 25 mIOU scores respectively. Can you let me know what could be a possible reason? Are the results reported on 1024*512 resolution images?

About the Table 5 in the paper

Hello, @Attila94 @oravus,I really appreciate for your work :)
The "Table 5: Recall@1 for VPR using different feature pooling types on Gardens Point (GP) and Tokyo 24/7 dataset" looks good.Is it convenient to provide the relevant implementation code? Of course, detailed guidance would be better.
I am always looking forward to your kind response.
Best regards.

HI, i have a little question about the Fig. 6 (Edge map) in the paper

What a wonderful work!!! And, I am curious about the edge map. How can I get this image, just like the below one in the paper?

Can you provide trained models?

CIConv is really interesting! Thanks a lot for your great work.
I wonder if you can release trained W-* models for the four experiments (Synthetic classification, CODaN classification, Semantic segmentation, Visual place recognition)?

NighttimeDriving test set labels

Where can I find the Nighttime Driving dataset labels you corrected? Can you provide a file link if you upload it to the GitHub repository？ Thank you ~

Synthetic classification dataset

Thanks a lot for your great work.
But there are some mistakes here:
FileNotFoundError: [Errno 2] No such file or directory: 'train/02691156.npz'

Question about the dark zurich_test test file

Thank you for sharing your great code. But I have some questions about the dark zurich_test test file. Can you provide the prediction result generation code for the Dark Zurich test? My revised code predicts results that do not meet the official requirements. thank you very much!

I am getting 0.216 for Dark Zurich .

using python eval.py '/CIConv-main/experiments/3_segmentation/cityscapes_w.pth.tar' --invariant 'W'

Night time driving
Score Average : 0.412
But for DarkZurich
Score Average : 0.216

Although I am getting this warning, but I do not think that can be the issue
/lib/python3.7/site-packages/torchvision/transforms/functional.py:405: UserWarning: Argument interpolation should be of type InterpolationMode instead of int. Please, use InterpolationMode enum.
"Argument interpolation should be of type InterpolationMode instead of int. "
/lib/python3.7/site-packages/torchvision/transforms/functional.py:405: UserWarning: Argument interpolation should be of type InterpolationMode instead of int. Please, use InterpolationMode enum.
"Argument interpolation should be of type InterpolationMode instead of int. "
/lib/python3.7/site-packages/torchvision/transforms/functional.py:405: UserWarning: Argument interpolation should be of type InterpolationMode instead of int. Please, use InterpolationMode enum.
"Argument interpolation should be of type InterpolationMode instead of int. "
/lib/python3.7/site-packages/torchvision/transforms/functional.py:405: UserWarning: Argument interpolation should be of type InterpolationMode instead of int. Please, use InterpolationMode enum.
"Argument interpolation should be of type InterpolationMode instead of int. "

how w_resnet101 weights have been trained.

Hi,
thanks for your code sharing. I am not clear what is the difference between w_resnet101 weights and resnet101 weights ?

thanks

i have some problems in downloading model.zip.

In experiment 4, the second downloading link has something wrong.
When i running this commend: wget -cO - https://cloudstor.aarnet.edu.au/plus/s/kya3ZjF5N3WqLVL/download > ./data/models.zip,
the commend telling me 403：Forbidden.

When i opening that link, the webpage telling me "Access denied. Download of a folder is not allowed."

dark zurich_test test file

hi~ Can you provide the prediction result generation code of Dark Zurich test? The prediction result of my modified code does not meet the official requirements. Thank you so much!