AI Paper Atlas

Research map / 3D vision / NeRF and radiance fields

MatrixCity: A Large-scale City Dataset for City-scale Neural Rendering and Beyond

Yixuan Li, Lihan Jiang, Linning Xu, Yuanbo Xiangli, Zhenzhi Wang, Dahua Lin, Bo Dai

ICCV 2023 · 3D vision › NeRF and radiance fields · approach: Sparse-input NeRF regularization · 128 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
naively train aerial and street data jointly (ablation)MipNeRF-360 · MatrixCity fused dataPSNR: 23.05 -> 17.07 (street), aerial 28.37Hurt
naively train aerial and street data jointly (ablation)TensoRF · MatrixCity fused dataPSNR: 22.10 -> 21.44 (street), aerial 27.26Hurt
increase street capture density from 5.0 m to 1.0 m (ablation)Instant-NGP · MatrixCity streetPSNR: 21.436 -> 32.444Improved
increase street capture density from 5.0 m to 1.0 m (ablation)MipNeRF-360 · MatrixCity streetPSNR: 27.75 -> 31.858Improved
model whole Small City aerial data with one model instead of five block models (ablation)MipNeRF-360 · MatrixCity aerialPSNR: 25.73 -> 24.41Hurt
model whole Small City aerial data with one model instead of five block models (ablation)TensoRF · MatrixCity aerialPSNR: 25.13 -> 25.09No clear effect
model whole Small City aerial data with one model instead of five block models (ablation)Instant-NGP · MatrixCity aerialPSNR: 25.69 -> 25.46No clear effect
split street data into smaller blocks (1 to 16 blocks) (ablation)MipNeRF-360 · MatrixCity street Small CityPSNR: 22.00 -> 26.67Improved
increase model capacity from Small to Large at 16 blocks (ablation)TensoRF · MatrixCity street Small CityPSNR: 22.96 -> 24.24Improved
increase model capacity from Small to Large at 16 blocks (ablation)MipNeRF-360 · MatrixCity street Small CityPSNR: 24.64 -> 26.67Improved
use street data instead of aerial data for renderingall baselines · MatrixCityPSNR: aerial overall 25.73 vs street overall 23.24 (MipNeRF-360)Hurt
use NeRF-W to filter dynamic objects in street scenesNeRF-W · MatrixCity dynamic streetqualitativeHurt

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