how to run the code

[Gchinanty]

I have installed matcaffe. And I have downloaded pretrained models.
But after I run DBPN_main_rgb.m at the root of DBPN-caffe which I have directly downloaded from GitHub. I only get the following results:
Bi cubic: NaN , NaN , NaN , NaN
dbpn: NaN , NaN , NaN , NaN
What's the matter with what l did? And could you tell me how I can get the correct results?
Thanks!!!

[alterzero]

@Gchinanty which test set did you use? is it jpg file?

[ynieWXJB]

I used png file as input. But I got the same results with Gchinanty.

[Gchinanty]

I used Set5 and it is *.png file just as your code showed. So I am very puzzled.
And now I have another question: what does 'train_4x_div_128_rgb.txt' mean in the 'DBPN_net_4x.prototxt'? Does it mean the input size? If not, could you tell me what is the input size of the net_4x?
Thanks a lot! @alterzero

[alterzero]

I'm not sure what's the problem here.
Did you use 'DBPN_net_4x.prototxt' for the evaluation? not 'DBPN_mat_4x.prototxt'?
Can you run caffe.set_mode_cpu()? or caffe.set_mode_gpu()?

'train_4x_div_128_rgb.txt' is used for training step which contains the list of your hdf5 files, the LR image size is 32x32.