D_model
Jon-drugstore opened this issue · 4 comments
thanks for your good work, the D_model would be released for continue train? Another question: Copy T and R images to trainA1 and trainA2? meaning ? how to prepare, it's for initial？
Thank you for your questions. I am a little confused. Can you describe them more clearly？
Thank you for your questions. I am a little confused. Can you describe them more clearly？
Q1: In your readme file, I am confused with the describes, "Prepare some images for synthesis. Copy T and R images to trainA1 and trainA2 separately."
My understanding is the R images are used for initial in train phase, how to prepare the R images for training?
Q2: I want to resume train from your model, but the final_net_D.pth is not provided. The D model would be released? thanks for your reply!
The answer to Q1: Because we don't have enough real data (I and T) to train for SIRR. Before our training, we need to get enough synthetic images I by combining images T and R. This step is necessary for all research in SIRR.
The answer to Q2: Of course, we will release it soon. Thank you for your reminder.
The answer to Q1: Because we don't have enough real data (I and T) to train for SIRR. Before our training, we need to get enough synthetic images I by combining images T and R. This step is necessary for all research in SIRR.
The answer to Q2: Of course, we will release it soon. Thank you for your reminder.