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
362 stars 77 forks source link

I don't know why I could not reproduce the validation results. #33

Closed jichaofeng closed 2 years ago

jichaofeng commented 2 years ago

For the Car class, The validation results are: bev AP: 73.51 48.49 37.47 3D AP: 62.28 39.16 30.51 Do you computing the disparity map using the point cloud or openCV BlockMatching?

Owen-Liuyuxuan commented 2 years ago

The result from openCV blockMatching and the point cloud should be no different.

Based on my observation and my past experience, it could be that you did not run disparity_precompute after changing the split file. This is the result with no/wrong disparity map (according to the paper and my own experience).

jichaofeng commented 2 years ago

Thank you. Your suggestion help me a lot. I run the program again according to the training schedule, Now the problem is solved.

JensenGao commented 2 years ago

Thank you. Your suggestion help me a lot. I run the program again according to the training schedule, Now the problem is solved.

Which backbone do you use for yolostereo3D?

jichaofeng commented 2 years ago

resnet34

---- 回复的原邮件 ---- | 发件人 | Jensen @.> | | 日期 | 2022年01月04日 11:13 | | 收件人 | @.> | | 抄送至 | @.**@.> | | 主题 | Re: [Owen-Liuyuxuan/visualDet3D] I don't know why I could not reproduce the validation results. (#33) |

Thank you. Your suggestion help me a lot. I run the program again according to the training schedule, Now the problem is solved.

Which backbone do you use for yolostereo3D?

— Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android. You are receiving this because you authored the thread.Message ID: @.***>

JensenGao commented 2 years ago

resnet34 ---- 回复的原邮件 ---- | 发件人 | Jensen @.> | | 日期 | 2022年01月04日 11:13 | | 收件人 | @.> | | 抄送至 | @.**@.> | | 主题 | Re: [Owen-Liuyuxuan/visualDet3D] I don't know why I could not reproduce the validation results. (#33) | Thank you. Your suggestion help me a lot. I run the program again according to the training schedule, Now the problem is solved. Which backbone do you use for yolostereo3D? — Reply to this email directly, view it on GitHub, or unsubscribe. Triage notifications on the go with GitHub Mobile for iOS or Android. You are receiving this because you authored the thread.Message ID: @.***>

Thx