zjr2000/GVL

Generated captions from models

Closed this issue · 4 comments

Hello, thank you for sharing the code! Would you also be able to provide the predicted captions from your model? I'm interested in the dense and paragraph captioning predictions for the ActivityNet dataset as I want to use it for my own evaluation.

Thank you for your reply :)
I noticed that you are also the author for PDVC: https://arxiv.org/pdf/2108.07781.pdf
Do you happen to have the predicted captions for this as well? I'm interested in the ones that reproduce the results from Table 3 and 4 for both ground-truth and predicted proposals.

The original files are lost. But I believe it is easy to reproduce all results using the pretrained checkpoints at https://github.com/ttengwang/PDVC#performance

Okay thank you so much