{ "schema_version": "1.0", "generated_by": "catalog.py", "datasets": { "euroc_mav": { "physical_id": "euroc_mav", "configs": [ "euroc_mav_april", "euroc_mav_chess" ], "n_configs": 2, "collections_per_config": { "euroc_mav_april": 2, "euroc_mav_chess": 1 }, "n_cameras": [ 2 ], "shutter": [ "global" ], "camera_models_benchmarked": [ "radtan" ], "boards": [ "apriltag 6x6 t36h11 tag 0.088 m", "chessboard 7x6 square 0.06 m" ], "source_type": [ "rosbag_topic" ], "tracks": [ "camera-intrinsics", "simulation (vs-GT)" ], "title": "EuRoC MAV calibration bags", "source": "ETH Zurich ASL", "homepage": "https://projects.asl.ethz.ch/datasets/euroc-mav/", "license": "CC BY (attribution)", "cite": "Burri2016EuRoC", "camera_hw": "2x Aptina MT9V034 (Skybotix VI-Sensor), mono, ~11 cm baseline", "resolution_fps": "752x480 @ 20 Hz", "lens": "standard (~80 deg H)", "imus": "ADIS16448 @ 200 Hz", "hosted": "recipe", "notes": "Hardware cam-IMU sync; april + checkerboard calib bags. Also the simulation-track GT config (vs Kalibr-wiki reference).", "sensors": [ "vi-sensor" ] }, "hilti_2022": { "physical_id": "hilti_2022", "configs": [ "hilti_2022" ], "n_configs": 1, "collections_per_config": { "hilti_2022": 6 }, "n_cameras": [ 1, 2 ], "shutter": [ "global" ], "camera_models_benchmarked": [ "equidistant" ], "boards": [ "apriltag 7x12 t36h11 tag 0.15 m" ], "source_type": [ "rosbag_topic" ], "tracks": [ "camera-intrinsics" ], "title": "Hilti-Oxford 2022 calibration bag", "source": "Hilti SLAM Challenge 2022", "homepage": "https://hilti-challenge.com/dataset-2022.html", "license": "CC BY-NC-SA 3.0 (NonCommercial!)", "cite": "Zhang2023HiltiOxford", "camera_hw": "Sevensense Alphasense Core: 5x 0.4MP GS mono (Sony IMX287-class) \u2014 front stereo (11 cm) + 2 lateral + 1 up", "resolution_fps": "720x540 @ 40 Hz", "lens": "fisheye-ish wide (126x92 deg)", "imus": "Bosch BMI085 @ 400 Hz (FPGA-synced)", "hosted": "recipe", "notes": "One calib bag; our 6 collections = cam0..cam4 individually + the cam0_cam1 front stereo pair.", "sensors": [ "alphasense-core" ] }, "hilti_2023_handheld": { "physical_id": "hilti_2023_handheld", "configs": [ "hilti_2023_handheld" ], "n_configs": 1, "collections_per_config": { "hilti_2023_handheld": 6 }, "n_cameras": [ 1, 2 ], "shutter": [ "global" ], "camera_models_benchmarked": [ "equidistant" ], "boards": [ "apriltag 6x6 t36h11 tag 0.08 m" ], "source_type": [ "rosbag_topic" ], "tracks": [ "camera-intrinsics" ], "title": "Hilti 2023 handheld ('Phasma') calibration bag", "source": "Hilti SLAM Challenge 2023", "homepage": "https://hilti-challenge.com/dataset-2023.html", "license": "CC BY-NC-SA 3.0 (NonCommercial!)", "cite": "Zhang2023HiltiOxford", "camera_hw": "Sevensense Alphasense Core (same 5-cam head as 2022)", "resolution_fps": "720x540 @ 10 Hz (published rosbag rate)", "lens": "fisheye-ish wide (~126x92 deg)", "imus": "Bosch BMI085 @ 200 Hz (published rate)", "hosted": "recipe", "notes": "Handheld device; 6 collections = cam0..cam4 individually + the cam0_cam1 front stereo pair.", "sensors": [ "alphasense-core" ] }, "monado_msd": { "physical_id": "monado_msd", "configs": [ "tum_mic" ], "n_configs": 1, "collections_per_config": { "tum_mic": 6 }, "n_cameras": [ 2 ], "shutter": [ "global" ], "camera_models_benchmarked": [ "equidistant" ], "boards": [ "apriltag 6x6 t36h11 tag 0.03 m" ], "source_type": [ "euroc" ], "tracks": [ "camera-intrinsics" ], "title": "Monado SLAM Datasets \u2014 MIC camcalib sequences", "source": "Collabora", "homepage": "https://huggingface.co/datasets/collabora/monado-slam-datasets", "license": "CC BY 4.0", "cite": "deMayo2025Monado", "camera_hw": "Valve Index HMD front stereo pair (sensor p/n unpublished), grayscale", "resolution_fps": "2x 960x960 @ 54 Hz", "lens": "fisheye (~120 deg, unofficial)", "imus": "Valve Index 6-axis @ ~1000 Hz (not used: camera-only folders)", "hosted": "recipe", "notes": "Repo id 'tum_mic' is a misnomer (pending rename to monado_msd): MIC01..MIC11 camcalib image folders, Kalibr-style target.", "sensors": [ "valve-index" ] }, "mvis_rig": { "physical_id": "mvis_rig", "configs": [ "mvis_elp", "mvis_elp_single", "mvis_elp_t265", "mvis_t265", "mvis_t265_ds", "mvis_t265_equidistant", "mvis_t265_eucm", "mvis_t265_fov" ], "n_configs": 8, "collections_per_config": { "mvis_elp": 5, "mvis_elp_single": 5, "mvis_elp_t265": 5, "mvis_t265": 5, "mvis_t265_ds": 5, "mvis_t265_equidistant": 5, "mvis_t265_eucm": 5, "mvis_t265_fov": 5 }, "n_cameras": [ 1, 2 ], "shutter": [ "global", "rolling" ], "camera_models_benchmarked": [ "double_sphere", "equidistant", "eucm", "fov", "radtan" ], "boards": [ "apriltag 6x6 t36h11 tag 0.088 m" ], "source_type": [ "rosbag_topic" ], "tracks": [ "camera-intrinsics", "rolling-shutter" ], "title": "VI-rig 4 IMUs + 4 cams (ours)", "source": "ours (MVIS rig, Yang et al. IJRR 2024)", "homepage": "https://huggingface.co/datasets/calibench-vi/datasets", "license": "CC BY 4.0", "cite": "Yang2024MVIS + Calibench", "camera_hw": "ELP stereo module (2x RS, mono out) + Intel RealSense T265 (2x OV9282 GS fisheye, BMI055)", "resolution_fps": "ELP 2x 640x480 @ 25 Hz; T265 2x 848x800 @ 30 Hz", "lens": "ELP standard-wide; T265 ~163 deg fisheye", "imus": "MicroStrain GX3-25 @ 500 Hz, GX3-35 @ 100 Hz, Xsens MTi-100 @ 400 Hz, T265 BMI055 @ 200 Hz", "hosted": "hosted (virig-4imus-4cams)", "notes": "5 apriltag bags; the ELP rolling-shutter pair seeds the rolling-shutter track (mvis_elp*). Rates/resolutions probed from the bags 2026-06-11.", "sensors": [ "elp-960p2cam-v90", "t265", "gx3-25", "gx3-35", "mti-100" ] }, "penncosyvio": { "physical_id": "penncosyvio", "configs": [ "penncosyvio", "penncosyvio_ds", "penncosyvio_eucm", "penncosyvio_fov" ], "n_configs": 4, "collections_per_config": { "penncosyvio": 3, "penncosyvio_ds": 1, "penncosyvio_eucm": 1, "penncosyvio_fov": 1 }, "n_cameras": [ 1 ], "shutter": [ "global" ], "camera_models_benchmarked": [ "ds", "equidistant", "eucm", "fov" ], "boards": [ "chessboard 7x6 square 0.108 m" ], "source_type": [ "images" ], "tracks": [ "camera-intrinsics" ], "title": "PennCOSYVIO GoPro intrinsic calibration", "source": "UPenn (Daniilidis group)", "homepage": "https://daniilidis-group.github.io/penncosyvio/", "license": "no license stated \u2014 citation-only, no redistribution grant", "cite": "Pfrommer2017PennCOSYVIO", "camera_hw": "3x GoPro Hero 4 Black (rig cameras C1/C2/C3); we use the intrinsic_calib image sets", "resolution_fps": "1920x1080 @ 30 Hz", "lens": "wide (118x70 deg, not full fisheye)", "imus": "none used (rig's VI-Sensor ADIS16488 @ 200 Hz not part of these sets)", "hosted": "recipe", "notes": "Rolling shutter, but calibrated as GS stills here (checkerboard intrinsic sets c1/c2/c3).", "sensors": [ "gopro-hero4" ] }, "tum_double_sphere": { "physical_id": "tum_double_sphere", "configs": [ "tum_fisheye_bf2m2020s23_ds", "tum_fisheye_bf2m2020s23_equidistant", "tum_fisheye_bf2m2020s23_eucm", "tum_fisheye_bf2m2020s23_fov", "tum_fisheye_bf5m13720_ds", "tum_fisheye_bf5m13720_equidistant", "tum_fisheye_bf5m13720_eucm", "tum_fisheye_bf5m13720_fov", "tum_fisheye_bm2820_ds", "tum_fisheye_bm2820_equidistant", "tum_fisheye_bm2820_eucm", "tum_fisheye_bm2820_fov", "tum_fisheye_bm4018s118_ds", "tum_fisheye_bm4018s118_equidistant", "tum_fisheye_bm4018s118_eucm", "tum_fisheye_bm4018s118_fov", "tum_fisheye_gopro_ds", "tum_fisheye_gopro_equidistant", "tum_fisheye_gopro_eucm", "tum_fisheye_gopro_fov" ], "n_configs": 20, "collections_per_config": { "tum_fisheye_bf2m2020s23_ds": 3, "tum_fisheye_bf2m2020s23_equidistant": 3, "tum_fisheye_bf2m2020s23_eucm": 3, "tum_fisheye_bf2m2020s23_fov": 3, "tum_fisheye_bf5m13720_ds": 3, "tum_fisheye_bf5m13720_equidistant": 3, "tum_fisheye_bf5m13720_eucm": 3, "tum_fisheye_bf5m13720_fov": 3, "tum_fisheye_bm2820_ds": 3, "tum_fisheye_bm2820_equidistant": 3, "tum_fisheye_bm2820_eucm": 3, "tum_fisheye_bm2820_fov": 3, "tum_fisheye_bm4018s118_ds": 3, "tum_fisheye_bm4018s118_equidistant": 3, "tum_fisheye_bm4018s118_eucm": 3, "tum_fisheye_bm4018s118_fov": 3, "tum_fisheye_gopro_ds": 3, "tum_fisheye_gopro_equidistant": 3, "tum_fisheye_gopro_eucm": 3, "tum_fisheye_gopro_fov": 3 }, "n_cameras": [ 1 ], "shutter": [ "global" ], "camera_models_benchmarked": [ "double_sphere", "equidistant", "eucm", "fov" ], "boards": [ "apriltag 6x6 t36h11 tag 0.088 m" ], "source_type": [ "rosbag_topic" ], "tracks": [ "camera-intrinsics" ], "title": "Double Sphere paper calibration sequences", "source": "TUM CVG (Usenko, Demmel, Cremers 2018)", "homepage": "https://cvg.cit.tum.de/research/vslam/double-sphere", "license": "not specified by the dataset authors (verified 2026-06-14) \u2014 recipe-only, linked not re-hosted", "cite": "Usenko2018DoubleSphere", "camera_hw": "IDS uEye body (model not named in paper), 5 lens setups: Lensagon BF2M2020S23, BF5M13720, BM4018S118, BM2820, GoPro replacement lens", "resolution_fps": "1280x1024 (body native; not stated in paper)", "lens": "195 / 183 / 126 / 122 / 150 deg FOV per lens", "imus": "none (camera-only)", "hosted": "recipe", "notes": "Was the unidentified 'tum_fisheye_*' source \u2014 identified 2026-06-11. 5 lenses x 3 apriltag sequences; the 'GoPro' is a replacement lens on the same body, not a GoPro camera. The wide lenses are exactly what DS/EUCM/FOV models exist for.", "sensors": [ "ueye-ui3241le", "bf2m2020s23", "bf5m13720", "bm4018s118", "bm2820", "gopro-replacement-lens" ] }, "tum_rgbd": { "physical_id": "tum_rgbd", "configs": [ "tum_rgbd_bag" ], "n_configs": 1, "collections_per_config": { "tum_rgbd_bag": 9 }, "n_cameras": [ 1 ], "shutter": [ "rolling" ], "camera_models_benchmarked": [ "radtan" ], "boards": [ "chessboard 6x7 square 0.04 m; chessboard 6x8 square 0.025 m; chessboard 7x9 square 0.04 m" ], "source_type": [ "rosbag_topic" ], "tracks": [ "camera-intrinsics" ], "title": "TUM RGB-D calibration bags", "source": "TUM CVG", "homepage": "https://cvg.cit.tum.de/data/datasets/rgbd-dataset", "license": "CC BY 4.0", "cite": "Sturm2012RGBD", "camera_hw": "Microsoft Kinect v1 (fr1/fr2), Asus Xtion Pro Live (fr3); RGB + IR calib bags", "resolution_fps": "640x480 @ 30 Hz", "lens": "standard", "imus": "none used (Kinect accelerometer present in bags)", "hosted": "recipe", "notes": "9 calib bags across fr1/fr2/fr3: rgb, ir, and large-checkerboard variants.", "sensors": [ "kinect-v1", "xtion-pro-live" ] }, "tum_rs": { "physical_id": "tum_rs", "configs": [ "tum_rs", "tum_rs_raw" ], "n_configs": 2, "collections_per_config": { "tum_rs": 2, "tum_rs_raw": 2 }, "n_cameras": [ 2 ], "shutter": [ "global" ], "camera_models_benchmarked": [ "equidistant" ], "boards": [ "apriltag 6x6 t36h11 tag 0.088 m" ], "source_type": [ "rosbag_topic" ], "tracks": [ "camera-intrinsics", "rolling-shutter" ], "title": "TUM rolling-shutter dataset calib", "source": "TUM CVG", "homepage": "https://cvg.cit.tum.de/data/datasets/rolling-shutter-dataset", "license": "not specified by the dataset authors (verified 2026-06-14) \u2014 recipe-only, linked not re-hosted", "cite": "Schubert2018DSORS", "camera_hw": "2x IDS uEye UI-3241LE-M-GL, mono (left GS, right RS ~29.47 us/row readout)", "resolution_fps": "1280x1024 @ 20 Hz", "lens": "Lensagon BM4018S118 (~126 deg)", "imus": "Bosch BMI160 @ 200 Hz", "hosted": "recipe", "notes": "Purpose-built RS benchmark: same camera model both sides, the right one runs rolling shutter (~29.47 us/row) in the evaluation sequences. The dedicated calib bags we use here are recorded with BOTH cameras in GS mode (the dataset's own protocol), which is why the configs say GS; the RS readout itself is what the rolling-shutter track will measure.", "sensors": [ "ueye-ui3241le", "bm4018s118", "bmi160" ] }, "tum_vi": { "physical_id": "tum_vi", "configs": [ "tum_vi_1024_ds", "tum_vi_1024_equidistant", "tum_vi_1024_eucm", "tum_vi_1024_fov", "tum_vi_raw", "tum_vi_standard_1024", "tum_vi_standard_512" ], "n_configs": 7, "collections_per_config": { "tum_vi_1024_ds": 8, "tum_vi_1024_equidistant": 8, "tum_vi_1024_eucm": 8, "tum_vi_1024_fov": 8, "tum_vi_raw": 8, "tum_vi_standard_1024": 8, "tum_vi_standard_512": 8 }, "n_cameras": [ 2 ], "shutter": [ "global" ], "camera_models_benchmarked": [ "double_sphere", "equidistant", "eucm", "fov" ], "boards": [ "apriltag 6x6 t36h11 tag 0.088 m" ], "source_type": [ "rosbag_topic" ], "tracks": [ "camera-intrinsics" ], "title": "TUM-VI calibration bags", "source": "TUM CVG", "homepage": "https://cvg.cit.tum.de/data/datasets/visual-inertial-dataset", "license": "CC BY 4.0", "cite": "Schubert2018TUMVI", "camera_hw": "2x IDS uEye UI-3241LE-M-GL, mono", "resolution_fps": "1024x1024 @ 20 Hz (512x512 downsampled variant)", "lens": "fisheye Lensagon BF2M2020S23 (195 deg)", "imus": "Bosch BMI160 @ 200 Hz", "hosted": "recipe", "notes": "Hardware-synced; 16-bit linear images with vignette calibration; 8 calib bags (cam1..cam8).", "sensors": [ "ueye-ui3241le", "bf2m2020s23", "bmi160" ] }, "tum_vie": { "physical_id": "tum_vie", "configs": [ "tum_event" ], "n_configs": 1, "collections_per_config": { "tum_event": 4 }, "n_cameras": [ 2 ], "shutter": [ "global" ], "camera_models_benchmarked": [ "equidistant" ], "boards": [ "apriltag 6x6 t36h11 tag 0.088 m" ], "source_type": [ "images" ], "tracks": [ "camera-intrinsics" ], "title": "TUM-VIE calibration (frame cameras)", "source": "TUM CVG", "homepage": "https://cvg.cit.tum.de/data/datasets/visual-inertial-event-dataset", "license": "CC BY 4.0", "cite": "Klenk2021TUMVIE", "camera_hw": "2x IDS uEye UI-3241LE-M-GL, mono (the rig's 2x Prophesee GEN4-CD event cameras are not used here)", "resolution_fps": "1024x1024 @ 20 Hz", "lens": "Lensagon BM4018S118 (101x76 deg)", "imus": "Bosch BMI160 @ 200 Hz (not used: camera-only image folders)", "hosted": "recipe", "notes": "calibA/calibB apriltag sequences, extracted image folders; visual stereo baseline 10.86 cm.", "sensors": [ "ueye-ui3241le", "bm4018s118", "bmi160" ] }, "urban_fisheye": { "physical_id": "urban_fisheye", "configs": [ "fisheye1_ds", "fisheye1_eucm", "fisheye1_fov", "fisheye2_ds", "fisheye2_eucm", "fisheye2_fov", "fisheye_dataset", "gopro_ds", "gopro_eucm", "gopro_fov" ], "n_configs": 10, "collections_per_config": { "fisheye1_ds": 1, "fisheye1_eucm": 1, "fisheye1_fov": 1, "fisheye2_ds": 1, "fisheye2_eucm": 1, "fisheye2_fov": 1, "fisheye_dataset": 3, "gopro_ds": 1, "gopro_eucm": 1, "gopro_fov": 1 }, "n_cameras": [ 1 ], "shutter": [ "global" ], "camera_models_benchmarked": [ "ds", "equidistant", "eucm", "fov" ], "boards": [ "chessboard 6x8 square 0.0325 m", "chessboard 6x8 square 0.0325 m; chessboard 6x8 square 0.117 m", "chessboard 6x8 square 0.117 m" ], "source_type": [ "images" ], "tracks": [ "camera-intrinsics" ], "title": "ImprovedOcamCalib fisheye sets", "source": "KIT (Urban, Leitloff, Hinz 2015)", "homepage": "https://github.com/urbste/ImprovedOcamCalib", "license": "GPL v2 (copyleft!)", "cite": "Urban2015ImprovedOcamCalib", "camera_hw": "Fisheye1: Lensation BF2M15520 (f=1.55mm) on GigE camera; Fisheye2: Lensation BF2M12520 (f=1.25mm) on VRmagic USB; GoPro: GoPro Hero (model unstated)", "resolution_fps": "still checkerboard images (resolutions unstated)", "lens": "~185 deg (Fisheye1/2), GoPro wide", "imus": "none (camera-only stills)", "hosted": "recipe", "notes": "Three DIFFERENT physical cameras \u2014 the combined 'fisheye_dataset' config treats them as collections, so cross-collection repeatability is invalid there (excluded from repeat_intrinsics_px).", "sensors": [ "bf2m15520", "bf2m12520", "gopro-hero" ] }, "virig": { "physical_id": "virig", "configs": [ "virig_flir", "virig_flir_t265", "virig_t265" ], "n_configs": 3, "collections_per_config": { "virig_flir": 5, "virig_flir_t265": 5, "virig_t265": 5 }, "n_cameras": [ 1, 2 ], "shutter": [ "global" ], "camera_models_benchmarked": [ "equidistant", "radtan" ], "boards": [ "apriltag 6x6 t36h11 tag 0.088 m" ], "source_type": [ "rosbag_topic" ], "tracks": [ "camera-intrinsics" ], "title": "VI-rig 4 IMUs + 3 cams (ours)", "source": "ours (MVIS rig, Yang et al. IJRR 2024)", "homepage": "https://huggingface.co/datasets/calibench-vi/datasets", "license": "CC BY 4.0", "cite": "Yang2024MVIS + Calibench", "camera_hw": "FLIR Blackfly S BFS-U3-16S2M (Sony IMX273, GS, mono) + Intel RealSense T265 (2x OV9282 GS fisheye, BMI055)", "resolution_fps": "Blackfly 720x540 @ 30 Hz (2x2 binned from 1440x1080); T265 2x 848x800 @ 30 Hz", "lens": "Blackfly standard; T265 ~163 deg fisheye", "imus": "MicroStrain GX3-25 @ 500 Hz, GX3-35 @ 100 Hz, Xsens MTi-100 @ 400 Hz, T265 BMI055 @ 200 Hz", "hosted": "hosted (virig-4imus-3cams)", "notes": "5 apriltag bags; rates/resolutions probed from the bags 2026-06-11.", "sensors": [ "bfs-u3-16s2m", "t265", "gx3-25", "gx3-35", "mti-100" ] }, "virig_5imus_4cams": { "physical_id": "virig_5imus_4cams", "configs": [], "n_configs": 0, "collections_per_config": {}, "n_cameras": [ 4 ], "shutter": [ "global" ], "camera_models_benchmarked": [], "boards": [ "apriltag 6x6 t36h11 tag 0.088 m" ], "source_type": [], "tracks": [ "camera-intrinsics" ], "hosted_only": true, "title": "VI-rig 5 IMUs + 4 cams (ours)", "source": "ours (RPNG VI sensor rig, 5 IMUs)", "homepage": "https://huggingface.co/datasets/calibench-vi/datasets", "license": "CC BY 4.0", "cite": "Calibench", "camera_hw": "2x Intel RealSense T265 (each 2x OV9282 GS fisheye, BMI055) = 4 fisheye cams", "resolution_fps": "4x 848x800 @ 30 Hz (two T265 stereo pairs: t265_01, t265_03)", "lens": "T265 ~163 deg fisheye (x2 units)", "imus": "MicroStrain GX3-25 @ 500 Hz, GX3-35 @ 100 Hz, Xsens MTi @ 400 Hz, 2x T265 BMI055 @ 200 Hz (5 IMUs)", "imu_topics": "/gx3_25/data, /gx3_35/imu/data, /mti/sensor/imu_raw, /t265_01/imu, /t265_03/imu", "hosted": "hosted (virig-5imus-4cams)", "notes": "7 kalibr AprilGrid bags (40-64 s each, 27.84 GB); 5 IMUs + 4 cams (two T265 stereo fisheye pairs). Topics/rates probed from the bags 2026-06-20. No calibration results yet \u2014 uploaded for benchmarking.", "sensors": [ "t265", "gx3-25", "gx3-35", "mti-100" ] } } }