search

zjukg / Structure-CLIP

[Paper][AAAI2024]Structure-CLIP: Towards Scene Graph Knowledge to Enhance Multi-modal Structured Representations
https://arxiv.org/abs/2305.06152
116 stars 6 forks source link

About TextRank1 result and ImageRank1 result #13

Closed lucas0214 closed 6 months ago

lucas0214 commented 6 months ago

Why do my experiment results show that TextRank1 is closer to IR-R@1 in Table 1 in the paper, while ImageRank1 is closer to TR-R@1 in Table 1 in the paper? Or is that the way it should be? Hope to reply, thank you!

BigHyf commented 6 months ago

你好，非常感谢你指出这个问题，是我们在写论文表格的时候表头出现了一些小问题，应该将IR-R@1和TR-R@1换一下，我们后续会修正，所以你的结果应该没有问题。TextRank@1是41.2，ImageRank@1是55.6，这个recall的结果可以看arxiv版本的附录部分。