search

kregmi / cross-view-image-synthesis

[CVPR 2018] Cross-View Image Synthesis using Conditional GANs, [CVIU 2019] Cross-view image synthesis using geometry-guided conditional GANs
55 stars 11 forks source link

Could you share the segmentation maps of the Dayton dataset generated by RefineNet? #16

Closed happsky closed 5 years ago

happsky commented 6 years ago

Could you share the segmentation maps of the Dayton dataset generated by RefineNet?

kregmi commented 6 years ago

I used the the code from https://github.com/guosheng/refinenet to generate the segmentation maps. I used RefineNet model trained using ResNet-101 on Cityscapes. I hope this helps

happsky commented 6 years ago

Thank you for your response! Have you done some pre-processing operations before extracting segmentation maps?

kregmi commented 6 years ago

I resized the images to 256x256 before extracting the segmentation maps since the original image resolution is 354x354 and used the default settings of RefineNet.

happsky commented 6 years ago

Thank you for your help.

happsky commented 6 years ago

Is "demo_predict_mscale_cityscapes.m" or "demo_test_simple_city.m" you used?

kregmi commented 6 years ago

This is what I used: refinenet_test_example_cityscapes.m

happsky commented 6 years ago

Sorry, I did not fine refinenet_test_example_cityscapes.m file, can you provide the link to me?

kregmi commented 6 years ago

Seems like they have changed the code. Temporarily, I have shared the segmentation maps here for you. Hope this helps!

OnePunchMann commented 5 years ago

could you share the segmentation maps again, i really need them

kregmi commented 5 years ago

Please find them here.