Explaining Human Preferences via Metrics for Structured 3D Reconstruction

Jack Langerman, Denis Rozumny, Yuzhong Huang, Dmytro Mishkin

Abstract: ”What cannot be measured cannot be improved” while likely never uttered by Lord Kelvin, summarizes effectively the driving force behind this work. This paper presents a detailed discussion of automated metrics for evaluat- ing structured 3D reconstructions. Pitfalls of each met- ric are discussed, and an analysis through the lens of ex- pert 3D modelers’ preferences is presented. A set of sys- tematic ”unit tests” are proposed to empirically verify de- sirable properties, and context aware recommendations re- garding which metric to use depending on application are provided. Finally, a learned metric distilled from human expert judgments is proposed and analyzed. The source code is available at https://github.com/s23dr/ wireframe-metrics-iccv2025