UniCalib:Targetless LiDAR-camera Calibration via Probabilistic Flow on Unified Depth Representations
Автор: han15
Загружено: 2026-02-22
Просмотров: 6
Описание:
UniCalib is a targetless LiDAR–camera calibration method that formulates calibration as probabilistic depth flow estimation on unified depth representations.
By converting cross-modal alignment into depth-to-depth matching and introducing a Perceptually Weighted Sparse Flow (PWSF) loss, UniCalib effectively handles occlusions, dynamic objects, and unreliable observations. The method achieves accurate and robust calibration without using calibration targets.
This video presents an overview of the method and experimental results on multiple real-world datasets, including KITTI Odometry, KITTI Raw, KITTI-360, and Waymo.
Повторяем попытку...
Доступные форматы для скачивания:
Скачать видео
-
Информация по загрузке: