search

hlsheng1 / RDIoU

"Rethinking IoU-based Optimization for Single-stage 3D Object Detection", ECCV2022 accept!
MIT License
128 stars 12 forks source link

train+val #26

Open zhouzhenhua994 opened 1 year ago

zhouzhenhua994 commented 1 year ago

I had the pleasure of reading your RDIoU article in ECCV 2022 and have a question for you, are your results on the kitti val set in table1 trained on the train+val dataset together?

hlsheng1 commented 1 year ago

Of course NOT! val results are just trianed on train set.....

zhouzhenhua994 commented 1 year ago

Of course NOT! val results are just trianed on train set.....

Thank you for your answer, what is your epoch in the three categories of training?

hlsheng1 commented 1 year ago

Just like the default setting. We use 200 epochs for three-calss training on KITTI dataset.

zhouzhenhua994 commented 1 year ago

就像默认设置一样。我们使用 200 个 epochs 在 KITTI 数据集上进行三级训练。

thank you

zhouzhenhua994 commented 1 year ago

就像默认设置一样。我们使用 200 个 epochs 在 KITTI 数据集上进行三级训练。

Can your method be applied to the point_head method as well?