search

qc17-THU / DL-SR

Tensorflow/keras implementation for image transformation from low-resolution (LR) image to super-resolved one, including single wide-field (WF) image super-resolution prediction and SIM reconstruction.
MIT License
98 stars 26 forks source link

Questions about the training and test data acquisition. #17

Closed Luchixiang closed 1 year ago

Luchixiang commented 1 year ago

Hi! Thank you for the awesome data and code. Aftering reading your paper, I still have one question about the training and test data acquisition:

I see that there a SIM_gt.mrc file and a RawSIMData_gt.mrc file in BioSR dataset. What's the difference between these two files? Which one is used for the ground truth in the trianing?

Thanks

qc17-THU commented 1 year ago

Hi! The RawSIMData_gt.mrc file includes 9 (3 orientations × 3 phases) diffraction-limited raw SIM images, while the SIM_gt.mrc file includes one SIM SR image used for GT in the training process of DLSR models.

Luchixiang commented 1 year ago

So the SIM_gt.mrc is reconstructed using the RawSIMData_gt.mrc through conventional method?

qc17-THU commented 1 year ago

Actually, the RawSIMData_gt.mrc is the average of 3 sets of high SNR raw SIM images used for training denoising models, and the SIM_gt.mrc is the average of 3 SIM images reconstructed from these 3 sets of raw SIM images used for DLSR tasks.

Luchixiang commented 1 year ago

Do you mean RawSIMData_gt.mrc is the average of the three highest SNR raw SIM images like RawSIMData_level_12.mrc, RawSIMData_level_11.mrc, RawSIMData_level_10.mrc of the F-actin dataset?

qc17-THU commented 1 year ago

No, another 3 sets of high SNR data.

Luchixiang commented 1 year ago

Got it. Thank you!