Interior Point Benchmark

Accuracy and performance of interior point / point-on-surface implementations, measured in your browser.

Loading…

Each row is measured from a GeoJSON geometry to an [x, y] pair, so whatever conversion a library needs is counted as that library's cost. Every row runs the dataset twice and only the second pass is timed, so the order the rows run in does not move their numbers. Verification uses this project's TypeScript verifyInteriorPoint; rows 1–2 are therefore self-verification. A ′ on a library name marks the centroidFirstInteriorPoint row, which asks whether the centroid is already inside before falling back to the scanline.

Initial data: PLATEAU Hiroshima 2024 (MLIT), CC BY 4.0. Drop a .geojson file on the page to measure your own data.