Fillable Region Detection Using Textual and Spatial Context

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

The manual process of creating fillable regions in digital forms is tedious, time-consuming, and error-prone, and conventional computer vision techniques are deficient in accurately identifying and classifying these regions.

Innovation Solution

An object detector and a language model are used to analyze candidate fillable regions in conjunction with textual and spatial context to generate accurate fillable region data, reducing the need for manual adjustments and resource consumption.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If manual form field authoring is used, then fillable regions can be created, but the process is tedious, time-consuming, and error-prone

Engineering Contradiction:

Improveform authoring efficiencyVSAvoidtime for manual field creation

Core Design Contradiction:

ProductivityVSLoss of time

Solution Approach 1:

The system performs automatic form field authoring by analyzing the form image itself to identify fillable regions, field labels, and assign field types without requiring manual user input for each field, thereby enabling the system to serve itself in the authoring process

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The patent replaces manual mechanical operations (clicking, dragging, configuring fields) with an automated computer vision system that uses image analysis and machine learning models to automatically detect and create form fields

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Extent of automation

If conventional computer vision techniques are used to create fillable boxes, then automation is achieved, but accuracy is deficient leading to incorrect placements and classifications

Engineering Contradiction:

Improveautomatic fillable box creationVSAvoidaccuracy of fillable region identification

Core Design Contradiction:

Extent of automationVSMeasurement precision

Solution Approach 1:

The patent segments the form analysis into distinct components: detecting candidate fillable regions, identifying field labels, determining field types, and grouping related fields, with each segment handled by specialized processing steps that collectively improve overall accuracy

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces an intermediary language model that acts as a mediator between the object detector and the final form structure, using textual context from the form to refine and verify the accuracy of detected fillable regions and their classifications

Inventive Principle:

Principle #24Intermediary (Mediator)

3Manufacturing precision

If manual creation of fillable fields is performed, then field positioning can be controlled, but the process requires extensive user modifications

Engineering Contradiction:

Improvefield positioning accuracyVSAvoiduser modification requirements

Core Design Contradiction:

Manufacturing precisionVSEase of operation

Solution Approach 1:

The system uses feedback loops where the language model analyzes textual context and spatial relationships to verify and refine the positioning and classification of detected fillable regions, continuously improving accuracy until the form is ready for use with minimal user intervention

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS12614405B2Facilitating identification of fillable regions in a form

Publication Date: 2026.04.28 ADOBE INC

AI summary

Methods and systems are provided for facilitating identification of fillable regions and/or data associated therewith. In embodiments, a candidate fillable region indicating a region in a form that is a candidate for being fillable is obtained. Textual context indicating text from the form and spatial context indicating positions of the text within the form are also obtained. Fillable region data associated with the candidate fillable region is generated, via a machine learning model, using the candidate fillable region, the textual context, and the spatial context. Thereafter, a fillable form is generated using the fillable region data, the fillable form having one or more fillable regions for accepting input.