Automated ICAO/ISO Compliance Validator for ID Portraits

Abstract

Abstract
The manual verification of identity document portraits against international
standards remains a labor-intensive and error-prone process, limiting scalability in
high-volume digital identity workflows. This paper presents an automated, open-source
compliance validator that programmatically enforces ICAO Doc 9303 and ISO/IEC
19794-5 requirements for ID portrait photographs. The proposed system integrates
MediaPipe's modern Tasks API for robust face landmark extraction, geometric rulebased checks for head-to-frame ratio (70–80%), eye-line positioning (50–60% from
top), and head pose tolerance (±10°), alongside environmental assessments for
background uniformity and lighting consistency. A modular, thread-safe architecture
enables dual deployment: a stateless Flask REST API for system integration and a
cross-platform Tkinter graphical interface for interactive validation, both producing
structured JSON reports with per-check status, quantitative metrics, and actionable
recommendations. Evaluation on a curated dataset of 500 portrait images across three
difficulty tiers demonstrates high precision (0.95) and recall (0.93) for overall
compliance detection, with average end-to-end latency of 218 ± 24 ms on CPU
hardware. The validator reduces manual review overhead by >85% while maintaining
full auditability through ephemeral, privacy-preserving processing. By bridging
standards-aligned rule enforcement with production-ready deployment, this work
provides a reproducible, offline-capable foundation for scalable identity photography
workflows in resource-constrained environments.