Results using your code are different to the reported in the ICCV paper

I have tried to obtain similar results to the ones reported in the HBB table:

### Task2 - Horizontal Leaderboard
| Approaches | mAP | PL | BD | BR | GTF | SV | LV | SH | TC | BC | ST | SBF | RA | HA | SP | HC |
|------------|:---:|:--:|:--:|:--:|:---:|:--:|:--:|:--:|:--:|:--:|:--:|:---:|:--:|:--:|:--:|:--:|
|[R²CNN++]|**75.35**|**90.18**|**81.88**|**55.30**|73.29|72.09|77.65|78.06|**90.91**|**82.44**|**86.39**|**64.53**|**63.45**|75.77|**78.21**|**60.11**|

I am using the validation set instead of the testing set because of the test annotations have not been released yed ... Could you provide the results with the validation set? Because they are too much worse compared to the ones that you report in the table ...

Am I missing something? I didn't change anything in your eval.py code .... But the mAP results are really dissappointing. I would like to know if someone has obtained similar results to the ones that the authors report.

Best,
Roberto Valle