search

huangzehao / caffe-vdsr

A Caffe-based implementation of very deep convolution network for image super-resolution
MIT License
273 stars 134 forks source link

sr_psnr less than bicubic psnr #39

Closed ghost closed 7 years ago

ghost commented 7 years ago

I tried your Test code directly and run the Demo_SR_Conv. But I get the following result:

iter:1 Elapsed time is 0.805680 seconds. sr_psnr: 11.556457 dB bi_psnr: 24.038857 dB

And the sr_reconstructed image looks like this:

figure Any recommendation will be appreciated.

huangzehao commented 7 years ago

Have you compiled matconvnet?

ghost commented 7 years ago

Yeah, I compiled and run the test. It compiled successfully and passed all the tests.

ghost commented 7 years ago

At first I got this error: Warning: Name is nonexistent or not a directory: /home/Dharma/caffe-vdsr/Test/matconvnet/matlab/mex

In path (line 109) In addpath (line 88) In vl_setupnn (line 13) In run (line 96) In Demo_SR_Conv (line 4) iter:1 Attempt to execute SCRIPT vl_nnconv as a function: /home/Dharma/caffe-vdsr/Test/matconvnet/matlab/vl_nnconv.m

Error in VDSR_Matconvnet (line 15) convfea = vl_nnconv(im_y,weight{1},bias{1},'Pad',1);

Error in Demo_SR_Conv (line 42) im_h_y = VDSR_Matconvnet(im_l_y, model,up_scale,use_cascade);

Then I found from stack overflow That I need to put the mex file in the root directory. So i copied the mex folder from my matconvnet that I got from vfleat and put it inside Test/matconvnet/matlab. And I run the Demo_SR_Conv. It works but it gives the above result.

ghost commented 7 years ago

Sorry My mistake. I have compiled your matconvnet and it works now. Thank you.

huangzehao commented 7 years ago

It is ok. Maybe there are some differences between latest matconvnet and the early version. I will check it.

ghost commented 7 years ago

Seems like some problem with my machine. It works on my another machine but does not work on this.