AI Paper Atlas

Research map / Detection and segmentation / 3D detection and driving

Delving Into Localization Errors for Monocular 3D Object Detection

Xinzhu Ma, Yinmin Zhang, Dan Xu, Dongzhan Zhou, Shuai Yi, Haojie Li, Wanli Ouyang

CVPR 2021 · Detection and segmentation › 3D detection and driving · approach: Monocular 3D object detection · 262 citations (OpenAlex)

Check your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
supervise the coarse center with the projected 3D center (ablation)CenterNet-based monocular 3D detector · KITTI validation, Car, moderate3D AP40: 11.12 -> 12.65Improved
add the IoU-oriented size loss (ablation)projected-center detector · KITTI validation, Car, moderate3D AP40: 12.65 -> 12.97Improved
discard distant training samples beyond 60m (ablation)projected-center detector with IoU-oriented size loss · KITTI validation, Car, moderate3D AP40: 12.97 -> 13.66Improved
retain multitask 2D detection alongside 3D detection (ablation)projected-center detector · KITTI validation, Car, moderate3D AP40: 10.30 -> 13.04Improved
add offset refinement for misalignment between the 2D center and projected 3D center (ablation)CenterNet-based monocular 3D detector · KITTI validation, Car, moderate3D AP40Improved
discard training samples beyond 40m instead of using the baseline sampling (ablation)CenterNet-based monocular 3D detector · KITTI validation, Car, moderate3D AP40: 12.97 -> 11.25Hurt
use soft distance weighting with c=40, T=1 (ablation)CenterNet-based monocular 3D detector · KITTI validation, Car, moderate3D AP40: 12.97 -> 11.74Hurt
use soft distance weighting with c=60, T=1 (ablation)CenterNet-based monocular 3D detector · KITTI validation, Car, moderate3D AP40: 12.97 -> 13.54Improved
use soft distance weighting with c=60, T=5 (ablation)CenterNet-based monocular 3D detector · KITTI validation, Car, moderate3D AP40: 12.97 -> 13.03No clear effect
remove uncertainty modeling from depth estimation (ablation)CenterNet-based monocular 3D detector · KITTI validation, Car, moderate3D AP40: 8.45 -> 9.97Hurt
use Laplace rather than Gaussian uncertainty modeling (ablation)CenterNet-based monocular 3D detector · KITTI validation, Car, moderate3D AP40: 10.41 -> 11.12Improved
add 3D GIoU regularization (ablation)CenterNet-based monocular 3D detector · KITTI validation, Car, moderate3D AP40: 11.12 -> 10.17; 10.19; 8.48Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy