how can I use this to do texture image retrieval?

xchhuang / pytorch_sliced_wasserstein_loss

An unofficial PyTorch implementation of "A Sliced Wasserstein Loss for Neural Texture Synthesis" paper [CVPR 2021].

13 stars 1 forks source link

how can I use this to do texture image retrieval? #2

Open jeannotes opened 2 years ago

jeannotes commented 2 years ago

it seems that you are using batchsize=1, so everytime, the extractor is not the same?how could we do that?

xchhuang commented 2 years ago

Hi,

Thanks for your interest. If you fixed the projection directions, the extractor will be the same and can be used to extract image features. Then I think comparing features of different images should give you some similarity scores for image retrieval. I am not sure if I answer your question.