Question about 4-task adverse weather training/testing code

[wudizongshi]

Hi, thank you for your excellent work and for releasing the code.

I am trying to reproduce the 4-task adverse weather removal results reported in the paper, including Snow100K-S, Snow100K-L, Outdoor-Rain, and RainDrop. However, I could not find the corresponding training and testing scripts in the repository.

Could you please let me know whether the training/testing code for this setting will be released? If so, may I ask when it is expected to be available?

Also, could you briefly share the training and evaluation protocol for this setting, such as the number of training iterations, patch size, batch size, learning rate schedule.

Thank you very much!

[Amazingren]

Hi, thank you for your excellent work and for releasing the code.

I am trying to reproduce the 4-task adverse weather removal results reported in the paper, including Snow100K-S, Snow100K-L, Outdoor-Rain, and RainDrop. However, I could not find the corresponding training and testing scripts in the repository.

Could you please let me know whether the training/testing code for this setting will be released? If so, may I ask when it is expected to be available?

Also, could you briefly share the training and evaluation protocol for this setting, such as the number of training iterations, patch size, batch size, learning rate schedule.

Thank you very much!

Hi @wudizongshi , thank you for your interest. Regarding the 4-task adverse weather removal. We directly adopt Histormer[1]'s code base (https://github.com/sunshangquan/Histoformer ) and replace the model with MIRAGE model.
Regarding the training iterations, patch size, batch size, etc, we kept all the same as Histormer.

We will work on it and include more detailed descriptions next week within our repo.

[1] Restoring images in adverse weather conditions via histogram transformer. ECCV2024