amazon-science / progressive-coordinate-transforms

Progressive Coordinate Transforms for Monocular 3D Object Detection, NeurIPS 2021
Apache License 2.0
64 stars 10 forks source link

Dose the network use depth pre-training? #11

Closed Cc-Hy closed 3 years ago

Cc-Hy commented 3 years ago

Hello, good job! I'd like to know if the network is pre-trained on a depth dataset? Thanks.

Willy0919 commented 3 years ago

Hello, good job! I'd like to know if the network is pre-trained on a depth dataset? Thanks.

For KITTI dataset, we use the pre-trained depth estimator DORN as in other methods, which is pre-trained on the KITTI depth dataset. For Waymo dataset, we pre-train the depth estimator based on the sparse depth maps generated by corresponding point clouds.

Cc-Hy commented 3 years ago

Thanks a lot.