UArizona/xcalib-a9

xcalib A9 r02_s01 HDF5 caches

Precomputed matching caches for camera-LiDAR cross-modal matching and targetless extrinsic calibration, built from the TUM Traffic / A9 s110 intersection recording. Each HDF5 file stores per-frame camera images, LiDAR point clouds, object detections, and the camera-LiDAR match matrices used to train and evaluate the xcalib matcher front-end.

This dataset repo is public and accompanies the accepted xcalib paper.

Dataset details

• Task: associate 2D camera detections with 3D LiDAR detections without calibration targets, then recover the camera-LiDAR extrinsic [R|t] (P = K [R|t]) from confident matches.
• Source: derived from the TUM Traffic / A9 dataset (s110 intersection).
• Sensors (per frame): two Basler cameras (s110_camera_basler_south1_8mm, s110_camera_basler_south2_8mm) and one LiDAR. Images are JPEG-encoded (decoded BGR to RGB); point clouds are XYZ in meters, in the same coordinate frame as the 3D boxes. Camera intrinsics are assumed known; xcalib solves only the extrinsic [R|t].

Splits

• a9_r02_s01_train.h5 - training
• a9_r02_s01_val.h5 - validation
• a9_r02_s01_test.h5 - held-out test (29 frames; never seen during training)

HDF5 schema

path	dtype	description
images/<camera>/data	uint8 (JPEG bytes) [N]	encoded RGB frames, one per frame
point_clouds/<lidar>/<frame>/xyz	float32 [P, 3]	LiDAR points, meters
labels/<sensor>/<frame>/num_camera_detections	int	number of 2D boxes K
labels/<sensor>/<frame>/num_lidar_detections	int	number of 3D boxes M
labels/<sensor>/<frame>/camera_bbox_2d	float32 [K, 4]	2D boxes (x1, y1, x2, y2) in pixels
labels/<sensor>/<frame>/lidar_bbox_3d	float32 [M, 6]	3D LiDAR boxes (6 values)
labels/<sensor>/<frame>/match_matrix	uint8 [K, M]	camera-LiDAR match labels
labels/<sensor>/<frame>/camera_names	bytes [K]	per-detection source camera (optional)
calibration	n/a	optional; absent in matching-only caches

Bounding-box arrays are allocated with capacity and sliced by the num_* counts ([0:K) / [0:M)).

Usage

from xcalib import load_dataset

loader = load_dataset("a9_dataset_r02_s01", split="test")
frame = loader[0]   # .image / .point_cloud / .bboxes_2d / .bboxes_3d / .match_matrix

Install with pip install xcalib; the loader fetches the cache from the Hub on first use and caches it locally.

License & upstream dataset

Cache files are released under CC BY-NC-ND 4.0. They derive from the TUM Traffic / A9 dataset; users must follow the upstream dataset's license terms. The xcalib source code is Apache-2.0.

Upstream citation

If you use these caches, please also cite the upstream A9 Intersection Dataset:

@inproceedings{zimmer2023a9intersection,
  title     = {A9 Intersection Dataset: All You Need for Urban 3D Camera-LiDAR Roadside Perception},
  author    = {Zimmer, Walter and Cre{\ss}, Christian and Nguyen, Hieu and Knoll, Alois C.},
  booktitle = {2023 IEEE 26th International Conference on Intelligent Transportation Systems (ITSC)},
  year      = {2023},
  note      = {arXiv:2306.09266},
}

Authors

Lihao Guo, Jiahao Tang, Tam Bang, Tianya Zhang, Austin Harris, Mina Sartipi, Siyang Cao

Citation

@article{guo2026xcalib,
  author  = {Guo, Lihao and Tang, Jiahao and Bang, Tam and Zhang, Tianya and
             Harris, Austin and Sartipi, Mina and Cao, Siyang},
  title   = {Position Encoding in Detection-Based LiDAR--Camera Matching:
             A Diagnostic Study at Infrastructure Sites},
  journal = {IEEE Sensors Letters},
  year    = {2026},
  note    = {Accepted. Paper URL pending. Code:
             https://github.com/radar-lab/xcalib},
}

Files

split	file	sha256
train	a9_r02_s01_train.h5	1b98298911fcd9e64f8e1651d7a8165c23f9e4df74fe957cec4caf734c13ab09
val	a9_r02_s01_val.h5	ba724986d6aef62e5834de97954098d4d3d8c0620bbb2ba82d271301f45ffbfa
test	a9_r02_s01_test.h5	bab6e7527682549bd3c01c3f0630365692f1e8906361d19fa5842bf0393739de