search

ZrrSkywalker / MonoDETR

[ICCV 2023] The first DETR model for monocular 3D object detection with depth-guided transformer
347 stars 29 forks source link

Submit question of KITTI benchmark. #12

Open Senwang98 opened 2 years ago

Senwang98 commented 2 years ago

Hi, @ZrrSkywalker I am trying to submit results to KITTI official. This is my first time to submit, so I am not sure my process is right?

  1. I think the final submit version is trained on trainval.txt.
  2. I am not sure the writelist is ['Car'] or ['Pedestrian', 'Car', 'Cyclist']? Because I found your submit only includes Car: image
ZrrSkywalker commented 2 years ago

Hi, thanks for your interest! We haven't release the test config yet. The submit version is trained on the full 'trainval.txt' and has less data augmentations as illustrated in the paper. Our result is only trained by 'Car' category for test set. We will complete the code after paper acceptance.

shanguangcv commented 2 years ago

When will the code of test config be updated?

yangfan293 commented 1 year ago

Hi, @ZrrSkywalker I am trying to submit results to KITTI official. This is my first time to submit, so I am not sure my process is right?

  1. I think the final submit version is trained on trainval.txt.
  2. I am not sure the writelist is ['Car'] or ['Pedestrian', 'Car', 'Cyclist']? Because I found your submit only includes Car:
image

Hello, may I ask what is the val set after the training set is set as trainval for the result you submitted? How do you find the best trained eopch? Also, are there any other Settings other than trainval? Because I used trainval for training, the test submitted result was only 13.+%. Hope I can get your help,thanks!

Hyeonjeong37 commented 8 months ago

Hello, how did you set the val set when training the model by using trainval?