A Generative AI Framework for Intelligent Flowchart Understanding and Automated Documentation
- Version
- Download 0
- File Size 356.00 KB
- File Count 1
- Create Date 24 March 2026
- Last Updated 24 March 2026
A Generative AI Framework for Intelligent Flowchart Understanding and Automated Documentation
Dr. K. Satyam1, Tupakula Sai2
1Associate Professor, Department of MCA, Annamacharya Institute of Technology & Sciences, Tirupati, AndhraPradesh, India.
2 Post Graduate, Department of MCA, Annamacharya Institute of Technology & Sciences, Tirupati, AndhraPradesh, India.
Abstract:Because they provide a visual representation of logical workflows and decision-making processes, flowcharts are essential in software development, system modelling, and instructional documentation. However, because of their intricate diagram structures, diverse shapes, embedded text, and irregular formatting styles, automatic flowchart interpretation is still a difficult task. Conventional rule-based and OCR-only methods sometimes fall short of simultaneously capturing semantic meaning and structural links. This paper introduces GenFlowchart, a Generative AI-powered system for automated documentation and intelligent flowchart comprehension. The suggested system combines geometric segmentation utilising the Segment Anything Model (SAM) to identify structural elements like processes, decisions, and connectors, optical character recognition (OCR) for text extraction, and picture preprocessing approaches. GPT-3.5 Turbo is then used to process the retrieved textual and spatial data using thoughtfully designed prompts, producing logical, easily comprehensible step-by-step flowchart descriptions. GenFlowchart performs better than traditional baseline models in terms of semantic accuracy, structural consistency, and user satisfaction measures, according to experimental evaluation on a variety of flowchart datasets. Significant gains in contextual accuracy and interpretation quality are shown by BERTScore and human evaluation scores. In addition to improving automatic diagram comprehension, the suggested approach offers a scalable basis for software analysis platforms, educational resources, and intelligent documentation systems.
Keywords:Generative Artificial Intelligence, Flowchart Understanding, Optical Character Recognition (OCR), Segment Anything Model (SAM), GPT-3.5 Turbo, Prompt Engineering, Diagram Interpretation, Intelligent Documentation, Computer Vision, NaturalLanguage Generation.