Subject-Level Unknown-Identity Identification from Leap Motion Controller 2 Hand Landmarks

2026-06-22 · Source: Takara TLDR - Daily AI Papers · Field: Technology & Digital — Artificial Intelligence & Machine Learning · Depth: Expert, medium

Summary

This work investigates subject recognition using Leap Motion Controller 2 (LMC2) hand landmark data under a subject-level unknown-identity identification protocol on the Multi View Leap2 Hand Pose (ML2HP) dataset. Researchers enriched the original geometric representation with fingertip-to-palm distances and palm-normalized inter-finger angular descriptors. Evaluation employed a Leave-One-Subject-Out (LOSO) protocol, where one subject was excluded from the enrolled set and treated as unknown, with an inner validation step for unknown-rejection threshold selection. The study compared an Extra Trees ensemble baseline against a learned embedding baseline and an MLP+OpenMax model. Extra Trees emerged as the strongest overall method, highlighting that the primary challenge lies in robust score separation between known and unknown probes, rather than just enrolled-subject discrimination. The findings support the viability of compact, interpretable landmark-based descriptors for contactless hand-based unknown-subject rejection and identification on small datasets.

Key takeaway

For Machine Learning Engineers developing contactless biometric systems, this research suggests focusing on robust score separation between known and unknown subjects. You should consider Extra Trees as a strong baseline for unknown-identity identification from hand landmarks, especially when working with small datasets. Incorporating enriched geometric features like fingertip-to-palm distances and palm-normalized inter-finger angles can significantly improve your model's performance and reliability for subject rejection.

Key insights

Contactless hand landmark data can enable robust unknown-subject identification and rejection, even with compact descriptors.

Principles

• Robust unknown-subject identification requires strong score separation.
• Geometric and angular hand descriptors enhance recognition.
• Leave-One-Subject-Out protocol is effective for unknown-identity evaluation.

Method

The method involves enriching LMC2 hand landmark data with fingertip-to-palm distances and palm-normalized inter-finger angular descriptors, then evaluating with LOSO and inner validation for thresholding.

In practice

• Use Extra Trees for strong known/unknown probe separation.
• Incorporate fingertip-to-palm and inter-finger angular descriptors.
• Employ LOSO for rigorous unknown-identity testing.

Topics

• Leap Motion Controller 2
• Hand Landmark Recognition
• Unknown-Identity Identification
• Open-Set Recognition
• Biometric Systems
• Machine Learning Models

Best for: Research Scientist, AI Scientist, Machine Learning Engineer, Computer Vision Engineer

Related on AIssential

• Person Identification from Contextual Motion — Person identification can be achieved interactively by analyzing motion responses to tailored visual cues, modeled probabilistically for high confidence.
• StableHand: Quality-Aware Flow Matching for World-Space Dual-Hand Motion Estimation from Egocentric Video — Accurate dual-hand motion estimation from egocentric video requires explicit per-frame observation quality awareness.
• UoU: A Universal Fingerprint Foundation Model Based on Large-Scale Unsupervised Learning — UoU reframes fingerprint feature extraction as a domain-specific foundation model problem using multi-level unsupervised learning.
• Train, Test, Re-evaluate: Schedule-Sensitive Evaluation of Generative Data for Hand Detection — Generative inpainting with multi-stage training improves hand detection performance and robustness to appearance variations.
• NeuroLip: An Event-driven Spatiotemporal Learning Framework for Cross-Scene Lip-Motion-based Visual Speaker Recognition — Event cameras and specialized processing can enable robust lip-motion-based speaker recognition across diverse scenes.
• Quality-Preserving Imperceptible Adversarial Attack on Skeleton-based Human Action Recognition — Adversarial attacks on S-HAR can be imperceptible by preserving motion quality through distribution-based optimization.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Takara TLDR - Daily AI Papers.