JMoonr / mmlane

multi-modal 3D Lane Detection
https://arxiv.org/abs/2209.05996
33 stars 0 forks source link

depth completion in the Depth-Aware Lifting #3

Open zengwz opened 1 year ago

zengwz commented 1 year ago

Hi, thanks for your nice work. In the Depth-Aware Lifting module, it is mentioned that you apply the depth completion as SFD. I check it and find that SFD uses TWISE to predict depth. However, the training code of TWISE is not released yet. Do you use the model trained by kitti dataset to predict depth of openlane dataset directly?