About Me
Affiliation
PhD Student at the Echizen Laboratory, National Institute of Informatics (NII) & Graduate University for Advanced Studies, SOKENDAI - Tokyo, Japan.
Research Focus
Working at the intersection of 3D computer vision, biometric security, and multimodal large vision-language models.
Core Problem
Developing privacy-preserving 3D face recognition systems that stay robust against deepfake and reconstruction attacks.
Mission
Building secure, irreversible biometric templates that balance high recognition accuracy with strong privacy guarantees.
Research Interests
▼
🧊3D Computer Vision
Analyzing and reconstructing 3D scenes from images or sensors — including 3D face meshes, point clouds, and depth estimation for secure biometric applications.
▼
🔐Biometric Security & Authentication
Designing robust systems that verify identity using physiological traits while defending against spoofing, replay, and presentation attacks.
▼
🤖Multimodal Vision-Language Models
Combining visual and textual understanding in large foundation models (LVLMs) to enable richer, context-aware reasoning about images, faces, and 3D content.
▼
🛡️Privacy-Preserving Machine Learning
Building ML pipelines that protect sensitive data through cancelable biometrics, secure template transformation, and differential privacy techniques.
▼
🎭Deepfake Detection & Generation
Studying how synthetic faces are created (GANs, diffusion, 3D morphing) and developing detection models that expose manipulation artifacts in 2D and 3D.
▼
🌐Federated Learning
Training models collaboratively across decentralized devices without sharing raw data — critical for privacy-preserving biometric systems across institutions.
▼
👤3D Face Recognition
Recognizing individuals from 3D facial geometry (meshes, point clouds) rather than 2D images, achieving stronger liveness and anti-spoofing guarantees.
News
1st-authored paper accepted at WACV 2026 — GFT-GCN: Privacy-Preserving 3D Face Mesh Recognition with Spectral Diffusion. 🎉