search

Owen-Liuyuxuan / visualDet3D

Official Repo for Ground-aware Monocular 3D Object Detection for Autonomous Driving / YOLOStereo3D: A Step Back to 2D for Efficient Stereo 3D Detection
https://owen-liuyuxuan.github.io/papers_reading_sharing.github.io/3dDetection/GroundAwareConvultion/
Apache License 2.0
361 stars 76 forks source link

Pretrained weights #66

Closed GFENGG closed 1 year ago

GFENGG commented 1 year ago

Hello, can you provide a pretrained weight about depth estimation?

Owen-Liuyuxuan commented 1 year ago

Sorry for the late reply. I just trained a new model using current code (some errors may occur related to contiguous tensors, but I have to look deeper into that before making public commit).

Here is the result at validation:

mean mae : 1.3309486414196072
mean rmse : 2.9350815975989972
mean inverse mae : 0.00470681552990103
mean inverse rmse : 0.007207281043113217
mean log mae : 0.06445663321755596
mean log rmse : 0.1023890602669587
mean scale invariant log : 0.09383321789658851
mean abs relative : 0.0633444312927093
mean squared relative : 0.011379228718697884

The config and checkpoints:

http://gofile.me/4jm56/f3lCTDkce

GFENGG commented 1 year ago

Thank you for your time, this is helpful for me