Giters

titu1994 / Image-Super-Resolution

Implementation of Super Resolution CNN in Keras.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

validation issue

maabdelghaffar opened this issue · comments

commented

hello,
i got a quick question here,
from the epochs i have acceptable results although its not as good as what mentioned in the paper, however, i don't know why the val_PSNRLoss at last epoch is 24.xxx and when validating the model it drops to average 13.xxxx
Am i doing something wrong here ? maybe i am confused with the difference between the val psnr loss and the model validating

thanks

Epoch 50/50
24576/24576 [==============================] - 4532s - loss: 0.0042 - PSNRLoss: 26.9141 - val_loss: 0.0055 - val_PSNRLoss: 24.2536
Validating model
Validating images from path
validated image:
126058_255_RGB_part2.tif
13.5239350796
validated image:
126058_255_RGB_part4.tif
13.6472463608
validated image:
126058_255_RGB_part3.tif
14.0380191803
Average PSNR value of validation images is
13.7364002069

commented

Perhaps the validation images are not represented in the training set, therefore the model can't upscale them properly.

There can be many reasons for this. I suggest increasing training dataset images.