A curated list of recent (2023 – 2026) LiDAR–Camera extrinsic calibration papers, toolboxes, and resources.
Inspired by and complementary to Deephome/Awesome-LiDAR-Camera-Calibration — this repo focuses on the newest works (roughly 2023 → 2026) that are not yet indexed there, including foundation-model / SAM / NeRF / 3D Gaussian Splatting based calibration, BEV-based methods, and joint Camera–LiDAR–Radar / Event camera calibration.
- Surveys & Reviews
- Target-based Methods
- Targetless Methods
- Online / Continuous Calibration
- Multi-sensor Joint Calibration
- Toolboxes
- Contributing
| Year | Title | Venue | Links |
|---|---|---|---|
| 2024 | Survey of Extrinsic Calibration on LiDAR-Camera System for Intelligent Vehicle: Challenges, Approaches, and Trends | IEEE T-ITS | paper |
| 2024 | A Review of Deep Learning-Based LiDAR and Camera Extrinsic Calibration | Sensors (MDPI) | paper |
| 2025 | Camera, LiDAR, and IMU Spatiotemporal Calibration: Methodological Review and Research Perspectives | Sensors (MDPI) | paper |
| 2023 | Deep Learning for Camera Calibration and Beyond: A Survey | arXiv:2303.10559 | paper |
Calibration using dedicated targets (checkerboards, ArUco, custom 3D objects).
-
Extrinsic Calibration of Camera and LiDAR Systems With Three-Dimensional Towered Checkerboards (2024, Int. J. Intelligent Systems), Ren et al. paper
- 3D towered checkerboard (3TC) target designed for robust camera–LiDAR extrinsic estimation.
-
One target to align them all: LiDAR, RGB and event cameras extrinsic calibration for Autonomous Driving (2026, arXiv:2511.12291), paper
- Single multi-modal target that jointly calibrates event camera + RGB + LiDAR rigs.
-
A long-range LiDAR–camera extrinsic calibration method for rail transit (2025, Scientific Reports), paper
- Custom long-range calibration target / workflow tailored to railway environments.
-
P2O-Calib: Camera-LiDAR Calibration Using Point-Pair Spatial Occlusion Relationship (2023, arXiv:2311.02062), Zhu et al. paper
- Uses point-pair occlusion cues around target boundaries for accurate 6-DoF extrinsics.
Methods that do not require any physical calibration target. Grouped by the core cue they rely on.
Hand-eye style calibration from ego-motion / odometry.
-
MDPCalib: Automatic Target-Less Camera-LiDAR Calibration From Motion and Deep Point Correspondences (2024, arXiv:2404.17298), Petek et al. paper
- Combines visual + LiDAR odometry with learned 2D-pixel ↔ 3D-point correspondences in a single optimization; generalizes across car / quadruped / UAV platforms.
-
OA-LICalib: Observability-Aware Intrinsic and Extrinsic Calibration of the Rolling-Shutter Camera and LiDAR-IMU Sensor Suite (2023, IEEE T-RO), Lv et al. paper code
- Continuous-time B-spline trajectory model with observability-aware degeneracy handling.
-
Targetless Intrinsics and Extrinsic Calibration of Multiple LiDARs and Cameras with IMU using Continuous-Time Estimation (2025, arXiv:2501.02821), paper
- Continuous-time joint calibration of multiple LiDARs + cameras + IMU without targets.
Classical targetless methods based on edge alignment, mutual information, or intensity/reflectance cues.
-
SensorX2car: Sensors-to-Car Calibration for Autonomous Driving in Road Scenarios (2023, arXiv:2301.07279), Yan et al. paper code
- Uses road-scene structure (lanes, vanishing points, ground plane) to calibrate LiDAR / camera / radar to the vehicle frame.
-
General, Single-shot, Target-less, and Automatic LiDAR-Camera Extrinsic Calibration Toolbox (2023, ICRA / arXiv:2302.05094), Koide et al. paper code
- Widely-used ROS1/ROS2 toolbox — direct, single-shot, targetless calibration from a single scan + image pair.
-
Pixel-Level Extrinsic Self-Calibration of High-Resolution LiDAR and Camera in Targetless Environments (IEEE RAL, baseline; arXiv:2103.01627), Yuan et al. paper code
- Edge-alignment based targetless calibration (
livox_camera_calib) — the canonical baseline widely used in 2023–2025 work.
- Edge-alignment based targetless calibration (
End-to-end / regression / flow-based learning approaches.
-
CalibFormer: A Transformer-based Automatic LiDAR-Camera Calibration Network (2024, ICRA / arXiv:2311.15241), Xiao et al. paper
- Transformer with multi-layer cross-modal correlation; reports ~0.88 cm / 0.056° on KITTI.
-
UniCalib: Targetless LiDAR-Camera Calibration via Probabilistic Flow on Unified Depth Representations (2025, arXiv:2504.01416), paper
- Projects both modalities into unified dense depth; learns a probabilistic flow field with uncertainty.
-
BEVCalib: LiDAR-Camera Calibration via Geometry-Guided Bird's-Eye View Representations (2025, arXiv:2506.02587), paper
- First calibration method operating directly on BEV features from raw data.
-
What Really Matters for Learning-based LiDAR-Camera Calibration (2025, arXiv:2501.16969), paper
- Systematic ablation revisiting what actually drives accuracy in learning-based LCC.
-
RobustCalib: Robust LiDAR-Camera Extrinsic Calibration with Consistency Learning (2023, arXiv:2312.01085), paper
- Self-supervised consistency learning for robust extrinsics under sensor noise.
-
Online, Target-Free LiDAR-Camera Extrinsic Calibration via Cross-Modal Mask Matching (2024, IEEE T-IV / arXiv:2404.18083), Huang et al. paper
- Uses large vision-model masks as cross-modal correspondences for online calibration.
Recent wave of calibration methods that leverage foundation models or neural scene representations.
-
Calib-Anything: Zero-training LiDAR-Camera Extrinsic Calibration Method Using Segment Anything (2024, China Automation Congress / arXiv:2306.02656), Luo et al. paper code
- SAM masks gate 3D–2D correspondences; zero extra training, works on arbitrary scenes.
-
SAM4D: Segment Anything in Camera and LiDAR Streams (2025, ICCV / arXiv:2506.21547), Xu et al. paper project code
- Cross-modal promptable segmentation over camera + LiDAR streams — foundation for cross-modal correspondence / calibration.
-
3DGS-Calib: 3D Gaussian Splatting for Multimodal SpatioTemporal Calibration (2024, arXiv:2403.11577), Herau et al. paper
- Uses LiDAR points as Gaussian anchors; orders of magnitude faster than NeRF-based calibration with better accuracy.
-
Targetless LiDAR-Camera Calibration with Anchored 3D Gaussians (a.k.a. TLC-Calib) (2025, arXiv:2504.04597), paper
- Jointly optimizes sensor poses + neural-Gaussian scene, using reliable LiDAR points as anchor Gaussians.
-
Robust LiDAR-Camera Calibration with 2D Gaussian Splatting (2025, arXiv:2504.00525), paper
- 2DGS reconstruction from LiDAR with photometric refinement of extrinsics.
-
MOISST: Multi-modal Optimization of Implicit Scene for SpatioTemporal Calibration (2023, IROS / arXiv:2303.03056), Herau et al. paper
- NeRF-based joint optimization of poses, LiDAR–camera extrinsics, and time offset.
-
SOAC: Spatio-temporal Overlap-Aware Multi-Sensor Calibration using Neural Radiance Fields (2024, CVPR / arXiv:2311.15803), Herau et al. paper
- NeRF-based multi-camera + LiDAR calibration exploiting overlapping FoVs.
-
INF: Implicit Neural Fusion for LiDAR and Camera (2023, IROS / arXiv:2308.14414), Zhou et al. paper
- Implicit neural field for joint LiDAR–camera representation and extrinsic refinement.
Methods that maintain valid extrinsics at runtime / detect miscalibration.
-
CalibRefine: Deep Learning-Based Online Automatic Targetless LiDAR–Camera Calibration with Iterative and Attention-Driven Post-Refinement (2025, arXiv:2502.17648), paper
- Common-feature discriminator + coarse homography + iterative attention refinement, fully online.
-
EdO-LCEC: Environment-Driven Online LiDAR-Camera Extrinsic Calibration (2025, arXiv:2502.00801), paper
- First environment-driven online calibration — a scene discriminator estimates feature density and adapts extraction.
Calibration methods that go beyond a single LiDAR ↔ Camera pair, jointly handling additional modalities.
-
OA-LICalib (2023, IEEE T-RO) — see Motion-based.
-
Targetless Intrinsics and Extrinsic Calibration of Multiple LiDARs and Cameras with IMU using Continuous-Time Estimation (2025, arXiv:2501.02821), paper
- Continuous-time joint calibration of multi-LiDAR + multi-camera + IMU without targets.
-
CLRNet: Targetless Extrinsic Calibration for Camera, Lidar and 4D Radar Using Deep Learning (2026, arXiv:2603.15767), paper
- End-to-end joint / pairwise calibration across camera + LiDAR + 4D radar.
-
RLCNet: An end-to-end deep learning framework for simultaneous online calibration of LiDAR, RADAR, and Camera (2025, arXiv:2512.08262), paper
- Joint LiDAR + RADAR + camera calibration with minimal supervision, real-time capable.
- One target to align them all: LiDAR, RGB and event cameras extrinsic calibration for Autonomous Driving (2026, arXiv:2511.12291), paper
- Unified multi-modal target for event + RGB + LiDAR calibration.
| Toolbox | Description | Link |
|---|---|---|
| OpenCalib / SensorsCalibration | Full multi-sensor calibration toolbox (targeted & targetless). | github / paper |
| CalibAnything | SAM-based zero-training targetless LiDAR-camera calibration. | github |
| SensorX2car | Road-scene targetless sensor-to-car calibration. | github |
| direct_visual_lidar_calibration | General single-shot targetless LiDAR-camera calibration (ROS1/ROS2). | github |
| livox_camera_calib | Pixel-level targetless self-calibration (HKU-MARS). | github |
| OA-LICalib | Observability-aware LiDAR–IMU–Camera continuous-time calibration. | github |
| velo2cam_calibration | Circular-hole + ArUco target toolbox — long-standing baseline. | github |
Pull requests are very welcome! Please follow this entry template:
- **Title** (Year, Venue / arXiv:xxxx.xxxxx), Authors. [paper](URL) [code](URL)
- One-line summary of the contribution.When adding a paper, place it in the most appropriate section (or suggest a new one). If a paper spans categories (e.g., SAM + online), pick the section that best captures its primary contribution and cross-link from the other section.
This list is released under CC0-1.0 — feel free to reuse, remix, and redistribute.