search

JianqiangRen / AAMS

Official repository of the paper 'Attention-aware Multi-stroke Style Transfer' in CVPR 2019
https://sites.google.com/view/yuanyao/attention-aware-multi-stroke-style-transfer
MIT License
79 stars 14 forks source link

about the freeze_model.py #2

Closed hello-trouble closed 4 years ago

hello-trouble commented 5 years ago

hello, Thank you for you excellent paper of CVPR2019. I am puzzled about the freeze_model.py .Could it deal with batch images in our project.Both the freeze_model.py and test.py only transfer one image ? Thank you for your help!

hello-trouble commented 5 years ago

when I run the train.py, an error of "TypeError: 'numpy.float32' object is not iterable" occured, but I don't change the origin code , do you meet this problem ?

JianqiangRen commented 5 years ago

hello, Thank you for you excellent paper of CVPR2019. I am puzzled about the freeze_model.py .Could it deal with batch images in our project.Both the freeze_model.py and test.py only transfer one image ? Thank you for your help!

Yes, after frozen, the model could only handle one pair of images(content/style image) in one fowrard process, since there may be different on sizes among batch images

JianqiangRen commented 5 years ago

when I run the train.py, an error of "TypeError: 'numpy.float32' object is not iterable" occured, but I don't change the origin code , do you meet this problem ?

I haven't met the problem you mentioned, please check your env, our project requests Python (version 2.7) Tensorflow (>=1.4) Numpy Matplotlib