Out of memory? about meta-sr-pytorch HOT 3 CLOSED

whether I have a mistake for training and testing.

And I see get_patch in multiscalesrdata.py,

when training,I use script:

python main.py --model metardn --save metardn --ext sep --lr_decay 200 --epochs 1000 --n_GPUs 4 --batch_size 16

And I put DIV2K train images in the root directory : (bin HR LR_bicubic), getnerate the LR_bicubic in prepare_dataset

when testing, I use script:

./experiment/metardn/model/model_1000.pt --test_only --data_test DIV2K --dir_data ./benchmark --scale 1.5 --n_GPUs 1 --data_range 1-1 0 --batch_size 1

And I put DIV2K val images in the benchmark directory in root.

Thank you very much!

from meta-sr-pytorch.

Sir,

Is it because the resolution for the HR and LR is very big? It is normal to run main.py when I use Set5 for testing.

please help me to check this error with me, thank you very much!

from meta-sr-pytorch.

I think, I should split the input image into several sub-images with a shave, and then do the SR operations for each sub-image, and then , I can merge the result for each sub-image, finally, I will get the whole SR image.

Thank you, and I will close the question.

from meta-sr-pytorch.