AI Paper Atlas

Research map / 3D vision / 3D generation

LetsMap: Unsupervised Representation Learning for Label-Efficient Semantic BEV Mapping

Nikhil Gosala, Kürsat Petek, B Ravi Kiran, Senthil Yogamani, Paulo L. J. Drews-Jr, Wolfram Burgard, Abhinav Valada

ECCV 2024 · 3D vision › 3D generation · approach: 3D perception for autonomous driving · 6 citations (OpenAlex)

Read the paperCheck 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
add unsupervised pretraining before BEV finetuning (ablation)LetsMap · KITTI-360 (1% BEV labels)mIoU: 26.47 -> 28.96Improved
add unsupervised pretraining with 100% BEV labels (ablation)LetsMap · KITTI-360 (100% BEV labels)mIoU: 32.68 -> 32.88No clear effect
add geometric pathway (implicit field) to pretraining (ablation)LetsMap · KITTI-360 (1% BEV labels)mIoU: 26.47 -> 27.75Improved
add temporal masked autoencoder semantic pathway (ablation)LetsMap with geometric pathway · KITTI-360 (1% BEV labels)mIoU: 27.75 -> 28.96Improved
increase masking ratio to 75% (versus 0%, 25%, 50%) (ablation)T-MAE pretraining · KITTI-360 (1% BEV labels)mIoU: 27.75 -> 28.96Improved
increase masking ratio to 90% (ablation)T-MAE pretraining · KITTI-360 (1% BEV labels)mIoU: 28.96 -> 27.31Hurt
use LetsMap instead of fully supervised PoBEV at 1% BEV labelsBEV mapping · KITTI-360 (1% BEV labels)mIoU: 19.09 -> 28.96Improved
use unsupervised LetsMap on nuScenes versus fully supervised PoBEV at 100% labelsBEV mapping · nuScenesmIoU: 30.90 -> 27.66Hurt
LetsMap vs SkyEye with FV labels on dynamic classesBEV mapping · KITTI-360 car and truckIoU: Car 39.77 (PoBEV) vs 30.94Mixed

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 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