You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When I reproduced CondInst, the results were different from those provided in the paper.
I train the ResNet50-Condinst for 90K iterations with batch size 16 on 4 3090 GPUs.
The results were as follows:
AP 32.015 AP50 57.328 AP75 30.931
CAR 43.365 Ped 33.444 Cyc 19.237
According to my observation, there is no instance id in waymo dataset with bicycle and ped object.
The text was updated successfully, but these errors were encountered:
When I reproduced CondInst, the results were different from those provided in the paper.
I train the ResNet50-Condinst for 90K iterations with batch size 16 on 4 3090 GPUs.
The results were as follows:
AP 32.015 AP50 57.328 AP75 30.931
CAR 43.365 Ped 33.444 Cyc 19.237
According to my observation, there is no instance id in waymo dataset with bicycle and ped object.
The text was updated successfully, but these errors were encountered: