omkarthawakar / composed-video-retrieval Goto Github PK

Hi, thank you for this wonderful work. However, while reproducing this work I notice that the code of your proposed model in file src/models/blip_cir.py and the webvid_covr are incomplete. It would be wonderful if you could update the remaining code and dataloader. Again, thank you for your wonderful work.

Hi, thank you for this wonderful work.

I appreciate you providing the code.

I'm trying to run your code, but there is a problem with the provided generated description file.

In the code, the description is loaded from annotation files. However, the description file is separate from the annotation file, so I have to concatenate these two files for data loading.

However, there is no information about 'id' column in the provided generated description file. So I can not concatenate the provided generated description file with the annotation file.

the example is below.

• the provided generated description file (there is id column)
  

• annotation file (there is no id column)
  

How can I match these two files?
Could you explain this problem? It would be helpful to me.

It should be pip install -r requirement.txt

Hi,
thanks for this great work~
And I want to ask a question that in the paper, it said that the video description is generated by image caption model and the hallucination is inevitable. And what does "visual inputs within BLIP latent space" mean? I am confused about how hallucinations are filtered out.

Looking forward to your reply, thank you~