abhay2005gupta/palm-recognition-dataset

24,000 high-quality images from 2,000 diverse participants worldwide - smartphone palm recognition dataset for biometric authentication

Participants & Demographics

• 2,000 diverse participants from multiple countries
• Balanced gender representation
• 6+ ethnic groups: Black, South Asian, Caucasian, Arab/Middle Eastern, Hispanic, East Asian
• Age range: Under 20 to 50+ years
• Both right-handed and left-handed individuals

Image Capture

• Smartphone-based: 200+ different models (iOS and Android)
• Dual-camera: Both front-facing and back-facing cameras
• Multiple backgrounds: 3 variations per configuration
• Complete coverage: Both left and right hands
• 12 images per participant

Rich metadata included

• Format: JSON and CSV
• Demographics: Gender, ethnicity, birth year, profession
• Technical: Device model, camera type, handedness
• File mappings: Links to all 12 images per participant

Comparison With Public Palmprint Databases

Dataset	Subjects	Images	Capture Method
This dataset	2,000	24,000	200+ smartphones
Tongji Contactless	300	12,000	Custom device
PolyU v3.0	600	12,000	Custom device
IITD Touchless	230	2,601	Fixed setup
11K Hands	190	11,076	Single camera
MPD (DeepMPV)	200	16,000	Smartphones

Full version of dataset is availible for commercial usage - leave a request on our website Axonlabs to purchase the dataset 💰

Use cases

• Biometric Authentication: Train palm recognition systems for secure authentication in mobile apps, banking, and access control
• Cross-Device Testing: Test algorithm performance across 200+ different smartphone models and camera qualities
• Fairness Research: Evaluate and improve model accuracy across different ethnicities, ages, and genders
• Multi-Modal Biometrics: Combine palm recognition with face, fingerprint, or iris for enhanced security

Why This Dataset?

• 2-3x larger than comparable public datasets
• Real smartphone capture (not specialized scanners)
• Comprehensive demographic diversity
• Dual-camera data for robustness testing
• Rich metadata for fairness research