tensorlayer / SRGAN

Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network
https://github.com/tensorlayer/tensorlayerx
3.24k stars 813 forks source link

An explanation of how to set up VGG-19 #225

Closed tobylerone closed 3 years ago

tobylerone commented 3 years ago

OSError: Failed to interpret file 'models/vgg19.npy' as a pickle

Hello, since I am new to tensorflow and deep learning, some of these questions may appear simple, but after a couple of days trying to get the SRGAN model working on my machine I still haven't solved this issue. Several others seem to have had the same problem but none of the responses have properly cleared things up for me. Previous answers seem to place the blame on an incomplete download, but i've downloaded the same file from many different sources and have had the same issue. How can I overcome this problem?

I downloaded vgg19.npy from the source provided and placed it in a ./model directory, as well as in site-packages/tensorlayer/models/, since I wasn't sure where it is supposed to be placed. The vgg19.npy file is just a pointer using git-lfs to the larger file. Is this what we use, or is there some preprocessing necessary to download this file and store it on our local system?

Also all the installed package versions are in line with the requirements.txt.

Thank you for your time,

Toby

p.s. please see attached the full error message

Capture d’écran 2020-11-11 à 13 18 51

tobylerone commented 3 years ago

Will close since issue solved now. Was fixed by downloading the complete file from here: https://mega.nz/file/xZ8glS6J#MAnE91ND_WyfZ_8mvkuSa2YcA7q-1ehfSm-Q1fxOvvs