search

hli1221 / densefuse-pytorch

DenseFuse (TIP2019) - Pytorch 0.4.1
103 stars 29 forks source link

Which dataset have you used in densefuse-pytorch? #2

Open lordbutters opened 4 years ago

lordbutters commented 4 years ago

I find the code in move_files.py shows that you have used "KAIST Multispectral database" as the dataset , but in your paper and densefuse-tensorflow , you have used coco2014 as the dataset. So which dataset have you used in densefuse-pytorch project?

hli1221 commented 4 years ago

I find the code in move_files.py shows that you have used "KAIST Multispectral database" as the dataset , but in your paper and densefuse-tensorflow , you have used coco2014 as the dataset. So which dataset have you used in densefuse-pytorch project?

coco2014 was used in training phase to train the autoencoder network. It's the only dateset used in training stage. We chosen some RGB images from KAIST to test fusion performance. So, KAIST is not mandatory.

lordbutters commented 4 years ago

I find the code in move_files.py shows that you have used "KAIST Multispectral database" as the dataset , but in your paper and densefuse-tensorflow , you have used coco2014 as the dataset. So which dataset have you used in densefuse-pytorch project?

coco2014 was used in training phase to train the autoencoder network. It's the only dateset used in training stage. We chosen some RGB images from KAIST to test fusion performance. So, KAIST is not mandatory.

thank you for your explanation