High quality form field detection data improves document understanding
Mind the Gaps: A Curated Benchmark for Form Field Detection
Computer Vision and Pattern RecognitionDatabases
Summary
Detecting form fields in documents is tricky because they often rely on white space and layout instead of visible content. The authors found that existing datasets for training these systems have a lot of errors, making it hard to check progress accurately. They created a new, carefully checked dataset called mini-CommonForms to provide cleaner, more reliable examples. They also tested current methods on this dataset to see how well they perform. This work aims to help researchers and developers build better tools for automating forms and making them easier to use.
What this means in practice
- •For document automation teams: Improve automated form processing systems by training on a higher-quality dataset for detecting form fields more reliably.
- •For accessibility software developers: Enhance accessibility tools by using better benchmarks to detect and interact with form fields in documents.
Authors
Iheb Brini, Omar Moured, Hamza Gbada, Elisa Barney
Abstract
Form Field Detection (FFD) is a fundamental component of document understanding systems, enabling applications ranging from large-scale industrial digitization to accessible form interaction for automated analysis. Unlike conventional object detection tasks, FFD is inherently challenging because fields are often defined by layout structure and whitespace rather than visible foreground content. Existing large-scale datasets frequently rely on heuristic annotation pipelines, resulting in noisy and inconsistent labels that hinder reliable evaluation. In this work, we introduce mini-CommonForms, a carefully curated FFD benchmark with consistent, high-quality annotations, and present a detailed evaluation of state-of-the-art detection approaches. The benchmark is designed to support reproducible research in document automation and accessibility-oriented applications. Dataset and code are available at https://github.com/moured/mini-commonforms