jizhuochen/iphone_stereo
iPhone stereo captures Complete paired iPhone camera exports and offline stereo calibration results. The capture and processing code is in JizhuoChen/stereo_iphone. Each captures/ subfolder is one fixed two-phone mount. Its data/ directory contains a 25-photo ChArUco calibration pair and three subsequent recording pairs. Camera A is the iPhone 17 Pro; Camera B is the iPhone 17 Pro Max. The complete phone exports include RGB frames, LiDAR depth and confidence maps, ARKit poses… See the full description on the dataset page: https://huggingface.co/datasets/jizhuochen/iphone_stereo.
iPhone stereo captures
Complete paired iPhone camera exports and offline stereo calibration results. The capture and processing code is in JizhuoChen/stereo_iphone.
Each captures/ subfolder is one fixed two-phone mount. Its data/ directory contains a 25-photo ChArUco calibration pair and three subsequent recording pairs. Camera A is the iPhone 17 Pro; Camera B is the iPhone 17 Pro Max. The complete phone exports include RGB frames, LiDAR depth and confidence maps, ARKit poses, Bluetooth timing records, audio, and location logs.
Each batch's calibration/outputs/photo_set_.../ directory contains:
calibration.json: board-only fit and automated quality assessment.extrinsics_fixed_arkit_intrinsics.json: recommended provisional calibration forprocessing rgb-reconstruct.stereo_extrinsics.yamlandintrinsics_{A,B}.yaml: the same provisional geometry for other processing commands.validation_fixed_arkit.json: held-out photo-pair checks.README.md: batch-specific results and limits.
Each calibration/outputs/recording_.../ directory contains Bluetooth clock mapping and a paired-frame table. The cameras were linked by Bluetooth but their exposures were not hardware synchronized. The printed ChArUco board is treated as having 34.5 mm squares. Its image coverage was limited, so the calibrations assume zero lens distortion and have unverified image-edge accuracy. Review the per-batch validation reports before using metric depth.