ddshan/hand_object_detector

Questions about AP on H+O

Opened this issue · 4 comments

Hello,I run the model on the testset of 100K dataset. But handobj_100K AP on H+O can not reach 46.9, which is only about 41. Other APs can reach what showed in the table and the paper. Do I have anything wrong or miss anything?

The model I run was downloaded from paper provided.

hi Can you please share with us the pertained model you have? We are also trying this model, but looks like some weights are missing. Thanks

Hi, I downloaded the model from this project, which is the same as what it provides now. You can train it without using any pretrained model, the results are same as the model they provided. But I can't reach the results in the paper.

@caihaunqai I had the same problem. I tested the model faster_rcnn_1_8_132028.pth provided by the project, but the AP on H+O only about 41. Have you solved this problem yet?