🥇 Calibench — Calibration Benchmark Leaderboard

Camera & camera-IMU calibration across toolboxes, on the Calibench datasets. Each tab is a task track. Methods get one score per track: on every roster dataset, the method's metric is divided by the best method's metric (capped; unsolved datasets get the cap) and averaged — robustness and accuracy in a single fair number, with coverage and common-subset accuracy shown alongside. Sim = vs held-out ground truth (definitive); Real = GT-free metrics. Results are grouped in batches (evaluation run date) — the board uses each tool's latest batch; tick Compare batches to track a tool over time. The calibration parameters behind each row are stored and linked (calib ↗) and compared in the parameter-agreement panel.