AI Paper Atlas

Research map / 3D vision / 3D generation

Towards Viewpoint Robustness in Bird's Eye View Segmentation

Tzofi Klinghoffer, Jonah Philion, Wenzheng Chen, Or Litany, Zan Gojcic, Jungseock Joo, Ramesh Raskar, Sanja Fidler, Jose M. Alvarez

ICCV 2023 · 3D vision › 3D generation · approach: 3D perception for autonomous driving · 17 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
Train BEV model with 25-50% of data transformed to target viewpoint by NVSCVT · DRIVE Sim, pitch -10 degIoU: 0.014 -> 0.165Improved
Train BEV model with NVS-transformed data for +0.8 m height targetCVT · DRIVE Sim, height 0.8 mIoU: 0.003 -> 0.214Improved
Pass train extrinsics at inference (Source*) instead of target extrinsicsCVT source model · DRIVE Sim, pitch -10 degIoU: 0.014 -> 0.078Improved
Randomly rotate extrinsics and 3D box labels during training (Extr Aug)CVT · DRIVE Sim, pitch +5 degIoU: 0.016 -> 0.028Improved
Extrinsic augmentation at source viewpointCVT · DRIVE Sim, source viewpointIoU: 0.170 -> 0.155Hurt
Transform 100% of training data instead of 25-50% (ablation)NVS augmentation · DRIVE Sim target rigsIoUHurt
Increase transformed fraction up to 50% (ablation)NVS augmentation · DRIVE Sim target rigsIoUImproved
Add SSIM loss, minimum loss and automasking to Worldsheet (ablation)Worldsheet NVS · AV NVS, 1K imagesDepth L1: 0.00763 -> 0.00707Improved
Add lidar depth supervision (ablation)Worldsheet NVS · AV NVS, 1K imagesDepth L1: 0.00707 -> 0.00657Improved
Train on multiple viewpoints rather than only the target viewCVT · DRIVE Sim, mean over viewsIoU: 0.19 mean test IoU (0.206 train views, 0.178 other views)Improved
Change camera viewpoint at inference for a source-trained model (CARLA analysis)LSS and CVT · CARLAIoUHurt

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