search

CompVis / adaptive-style-transfer

source code for the ECCV18 paper A Style-Aware Content Loss for Real-time HD Style Transfer
https://compvis.github.io/adaptive-style-transfer/
GNU General Public License v3.0
723 stars 139 forks source link

Still runnable in Colab? #42

Closed galabyca closed 3 years ago

galabyca commented 3 years ago

I'm not able to run it on Colab.

From the command !CUDA_VISIBLE_DEVICES=0 python main.py \ --model_name=model_pollock \ --phase=inference \ --image_size=512 \ --ii_dir=./seed0168.png \ --save_dir=./out

I'm getting [...]

Start inference. [] Reading latest checkpoint from folder ./models/model_pollock/checkpoint. [] Reading latest checkpoint from folder ./models/model_pollock/checkpoint_long. Load checkpoint model10_pollock_bks10_flw100_300000.ckpt-300000. Initial step: 300000. INFO:tensorflow:Restoring parameters from ./models/model_pollock/checkpoint_long/model10_pollock_bks10_flw100_300000.ckpt-300000 I1031 15:09:23.577128 140008567285632 saver.py:1284] Restoring parameters from ./models/model_pollock/checkpoint_long/model10_pollock_bks10_flw100_300000.ckpt-300000 [*] Load SUCCESS 0it [00:00, ?it/s] Inference is finished.

Here is the Colab Notebook

m990130 commented 3 years ago

Hey, I think you should specify the folder containing all the content images path in --ii_dir instead of the image itself.