AIR-VIEW: The Aviation Image Repository for Visibility Estimation of Weather, A Dataset and Benchmark

2025-06-25

AI Summary

The aviation visibility estimation community has long suffered from a lack of publicly available, diverse, large-scale, real-world image datasets with standardized visibility annotations. Method: This work introduces AviVis—the first aviation-specific visibility image repository, constructed from the FAA’s meteorological camera network—comprising hundreds of thousands of images captured across multiple geographic locations and diverse weather conditions. Visibility labels are uniformly annotated per the latest ASTM E3092 standard, and a cross-dataset standardized evaluation benchmark is established. Contribution/Results: We systematically evaluate three state-of-the-art deep learning models alongside traditional baseline methods on AviVis and several public datasets, delivering reproducible benchmark performance reports. This work fills a critical data gap in aviation visual perception, establishes a standardized evaluation paradigm for visibility estimation, and provides an authoritative data foundation and performance reference for future research.

Technology Category

• Computer Vision: Interpretability, Explainability, and Transparency
• Application Domains: Mobility, Driving & Flight
• Machine Learning: Evaluation and Analysis

Application Category

• Security and Privacy: Data transparency and provenance
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Systems and Infrastructure for Web, Mobile and WoT: Web performance, measurement, and characterization

Abstract

Machine Learning for aviation weather is a growing area of research for providing low-cost alternatives for traditional, expensive weather sensors; however, in the area of atmospheric visibility estimation, publicly available datasets, tagged with visibility estimates, of distances relevant for aviation, of diverse locations, of sufficient size for use in supervised learning, are absent. This paper introduces a new dataset which represents the culmination of a year-long data collection campaign of images from the FAA weather camera network suitable for this purpose. We also present a benchmark when applying three commonly used approaches and a general-purpose baseline when trained and tested on three publicly available datasets, in addition to our own, when compared against a recently ratified ASTM standard.

Problem

• Lack of public datasets for aviation visibility estimation
• Need for diverse, large-scale tagged weather image data
• Absence of benchmarks for ML in visibility estimation

Innovation

• Introduces FAA weather camera dataset for visibility estimation
• Benchmarks three common ML approaches on datasets
• Compares results against ASTM standard for validation