Hi thx for your excellent work. Recently I just validate your provided trained-model and script. After I submit the result to kitti, the result of three difficult levels are: 90.51 %, 89.90 % and 80.68 % respectively. The moderate level result is close to your paper report(Table 3, 89.85%), however the hard level result is far from the expected one(80.68% vs 86.97%). So I wonder what's the experiment settings in your Table 4, also with ensemble model or not ? Can u help me to check, I just run your code without any modification. Thx
86.97% is the result of ensemble models.
As KITTI changed the evaluation rule after we submit our paper, the results you get may be different from the reported results in our paper.
Hi thx for your excellent work. Recently I just validate your provided trained-model and script. After I submit the result to kitti, the result of three difficult levels are: 90.51 %, 89.90 % and 80.68 % respectively. The moderate level result is close to your paper report(Table 3, 89.85%), however the hard level result is far from the expected one(80.68% vs 86.97%). So I wonder what's the experiment settings in your Table 4, also with ensemble model or not ? Can u help me to check, I just run your code without any modification. Thx