"PPCF (PALMPRINT-FINGERPRINT CONTACTLESS FUSION DATASET)"

"AbstractWe introduce PPCF, a contactless multimodal biometric dataset that pairs four\u2011finger \u201cfingerslap\u201d images with full\u2011palm images from the same individuals.\u202fData were collected from 40 adult volunteers (18\u202ffemale, 22\u202fmale; ages\u202f19\u201354) using two flagship smartphones under identical indoor lighting at a sensor distance of ~15\u202fcm.\u202fFor each participant, the iPhone\u202f13\u202fPro\u202fMax session produced 56 images (18 left\u2011finger, 18 right\u2011finger, 10 right\u2011palm, and 10 left\u2011palm), while the Google\u202fPixel\u202f7a session produced 54 images (17 left\u2011finger, 17 right\u2011finger, 10 right\u2011palm, and 10 left\u2011palm).\u202fIn total, the dataset contains 4,400 high\u2011resolution (1080\u202f\u00d7\u202f1920\u202fpx) RGB images stored as lossless PNG files.Every image is accompanied by (i) segmentation masks for individual fingertips and palm regions, (ii) quality scores\u2014NFIQ\u202f2.1 for fingerprints and both Local (LQA) and Global (GQA) Quality Assessments for palms, and (iii) structured metadata capturing device model, hand laterality, capture angles, ambient lux, and exposure parameters.\u202fWe provide subject\u2011disjoint train, development, and test splits (60\u202f\/\u202f20\u202f\/\u202f20\u202f%) as well as cross\u2011device subsets to facilitate domain\u2011shift research.CFPMB\u201140 fills a critical gap in publicly available resources by enabling reproducible experiments on feature\u2011level fusion of fingerprints and palmprints.\u202fBaseline Vision Transformer scripts achieve 96.3\u202f% identification accuracy\u2014surpassing unimodal ViT baselines and approaching commercial performance\u2014demonstrating the dataset\u2019s utility for advanced fusion studies.\u202fThe corpus is released on IEEE\u202fDataPort under a CC\u202fBY\u2011NC\u202f4.0 license; commercial use requires separate permission."