search

ChenDelong1999 / polite-flamingo

🦩 Visual Instruction Tuning with Polite Flamingo - training multi-modal LLMs to be both clever and polite! (AAAI-24 Oral)
https://arxiv.org/abs/2307.01003
63 stars 3 forks source link

All Image data #4

Closed liaoning97 closed 1 year ago

liaoning97 commented 1 year ago

This work is wonderful. However, when I downloading the datasest, I found that the released resized_images.zip in huggingface are not all the images corresponding to PF-1M.json. I wonder whether you plan to open all the images contained in PF-1M.json? Thanks~

ChenDelong1999 commented 1 year ago

Hi, thanks for your interest in our work 😀

Yes, the resized_images.zip only contains the image files that have been rendered (with colored boxes or arrows) or been renamed (from ELEVATOR classification dataset), otherwise it is impossible to align local images with PF-1M image paths.

Unfortunately, we cannot release other image data as we are not permitted to do so by the company, and unapproved large-scale data transfers could trigger alerts in the server system.

However, all images from our dataset are sourced from publicly available datasets, which should be straightforward to download. Furthermore, the file structure in our dataset is consistent with the structure after direct unzipping from the source datasets.

We extend our sincerest apologies for any inconvenience this may cause🙏. If you encounter any issues or have any questions during the process of aligning the datasets, please feel free to ask. We would be more than happy to assist you.

liaoning97 commented 1 year ago

Thank you very much for your kind reply. I will try to reproduce the dataset later.