about the reproduce results of Table 6

[ZHUXUHAN]

hI, dear sir, thanks for your provided project.
i just follow the scripts to reproduce the results of Table 6 (fully supervised), but the result is :
SGG eval: R @ 20: 0.0501; R @ 50: 0.0809; R @ 100: 0.1042; for mode=sgdet, type=Recall(Main).
SGG eval: ng-R @ 20: 0.0505; ng-R @ 50: 0.0827; ng-R @ 100: 0.1100; for mode=sgdet, type=No Graph Constraint Recall(Main).
SGG eval: zR @ 20: 0.0183; zR @ 50: 0.0339; zR @ 100: 0.0448; for mode=sgdet, type=Zero Shot Recall.SGG eval: ng-zR @ 20: 0.0028; ng-zR @ 50: 0.0048; ng-zR @ 100: 0.0097; for mode=sgdet, type=No Graph Constraint Zero Shot Recall.
SGG eval: mR @ 20: 0.0063; mR @ 50: 0.0105; mR @ 100: 0.0139; for mode=sgdet, type=Mean Recall.
There seems to be a discrepancy with the reported results.

[rafa-cxg]

HI 请问问题解决了吗？

[vineet2104]

@zyong812 Can you please address this issue? Even I'm unable to reproduce results in Table 6 after following all instructions. Metrics for object detection make sense, but those for SGG are very poor as mentioned in the original issue.