Currently available open-source repository only provides the images, without including any of the “golden clues” / “quality descriptions” mentioned in the paper.
Personally, I highly recognize this work as it offers very valuable perspectives. However, due to the lack of key content, it is impossible to carry out follow-up work based on this research.
Personally I do not understand why authors claiming that they need to avoid leaking the corpus. If the goal is to create a ranking list, open-sourcing the labels will better enhance fairness and credibility. Therefore, I hope the authors can make the dataset open-source to the community, so that the actual contributions of this paper can match its publication status.
Currently available open-source repository only provides the images, without including any of the “golden clues” / “quality descriptions” mentioned in the paper.
Personally, I highly recognize this work as it offers very valuable perspectives. However, due to the lack of key content, it is impossible to carry out follow-up work based on this research.
Personally I do not understand why authors claiming that they need to avoid leaking the corpus. If the goal is to create a ranking list, open-sourcing the labels will better enhance fairness and credibility. Therefore, I hope the authors can make the dataset open-source to the community, so that the actual contributions of this paper can match its publication status.