search

njucckevin / ADS-Cap

A Framework for Accurate and Diverse Stylized Captioning with Unpaired Stylistic Corpora
9 stars 2 forks source link

style_extract #3

Open ama6eus opened 6 months ago

ama6eus commented 6 months ago

Hi,@njucckevin: I have some problem in the style_extract module. I can not find the bert_style_classifier model,"save_every_head-flickr7k-val.pt" and "save_every_head-senticap-val.pt". Besides, I can not find the code of "Similar Scenes arouse Similar Emotions" about style phrase extractor on the internet. Could you provide these models for referrence? I'd appreciate your help!

njucckevin commented 6 months ago

Hi,

Sorry I forgot to upload these checkpoints before. Now you can download these checkpoints in here, and place the save_every_head-flickr7k-val.pt, save_every_head-senticap-val.pt and two pytorch_model.bin in flickr7k and senticap respectively.

This style extractor are from paper Similar scenes arouse similar emotions: Parallel data augmentation for stylized image captioning. They constructed a style classifier base on Bert, and use the attention score to extract the style phrase.

ama6eus commented 6 months ago

Hi,

Sorry I forgot to upload these checkpoints before. Now you can download these checkpoints in here, and place the save_every_head-flickr7k-val.pt, save_every_head-senticap-val.pt and two pytorch_model.bin in flickr7k and senticap respectively.

This style extractor are from paper Similar scenes arouse similar emotions: Parallel data augmentation for stylized image captioning. They constructed a style classifier base on Bert, and use the attention score to extract the style phrase.

Thanks for your reply!