International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 12 Issue: 03 | Mar 2025
p-ISSN: 2395-0072
www.irjet.net
Automated Font Recommendation System Using Deep Learning Asma Begum1, Dayala Akshitha2, Nalla Bharat Kumar3, Ms. V. Alekya4 1B-Tech 4th year, Dept. of CSE (DS), Institute of Aeronautical Engineering
2 B-Tech 4th year, Dept. of CSE (DS), Institute of Aeronautical Engineering 3 B-Tech 4th year, Dept. of CSE (DS), Institute of Aeronautical Engineering
4Assistant Professor, Dept. of CSE (DS), Institute of Aeronautical Engineering, Telangana, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - This project introduces an automated system
Understanding the Font Selection:
that helps users choose the best fonts for different types of documents, such as academic papers, business reports, legal documents, and creative works. The system uses a machine learning model called DenseNet-201 to suggest fonts based on the user’s needs and preferences. To develop this system, a collection of Google Fonts was organized into five categories: Sans-serif, Serif, Handwriting, Display/Decorative, and Monospaced. These fonts were then transformed into grayscale images, each 300x300 pixels, for training the DenseNet-201 model. After the model is trained, it analyzes input documents by extracting and classifying text, which helps it suggest the most suitable font style, size, and weight to improve readability and appearance. The system was tested with a dataset of fonts and proved to be effective at accurately recognizing different font styles and providing helpful recommendations for various document types. By automating the font selection and document analysis process, this system offers users a useful tool for improving their documents’ visual quality and professionalism.
Font Categories and Styles: Fonts are grouped into several categories, each serving distinct purposes based on the nature of the document. Serif fonts, for instance, are commonly used in formal documents such as academic and legal texts due to their traditional appearance, while Sansserif fonts are preferred for modern and digital content because of their clean and minimal design. Understanding these categories—Serif, Sans-serif, Script, Display, Monospaced, and Handwritten—helps in selecting a font that complements the tone and purpose of the document. Readability and Legibility: The primary goal of font selection is to ensure the text is easy to read. Fonts with clear and distinguishable characters enhance legibility, especially for long-form documents. Factors such as font size, weight, and spacing also play a significant role in improving readability. For instance, Sans-serif fonts are often used in digital documents due to their high legibility on screens, while Serif fonts are better suited for printed text, where they guide the reader’s eye across lines.
Key Words: Font recommendation, DenseNet-201, Google Fonts, image conversion, text extraction, classification, automated system.
Document Context and Audience: The context of the document and its intended audience are critical in font selection. Business reports, for example, demand professional, clean fonts like Sans-serif, whereas creative projects may benefit from more expressive fonts like Handwritten or Display. Understanding the preferences and expectations of the target audience helps in selecting a font that enhances the document’s message and visual appeal without distracting from its content.
1. INTRODUCTION In the digital age, font selection plays a critical role in enhancing the readability, aesthetic appeal, and professionalism of documents. Whether for academic, business, legal, or creative purposes, choosing the right font style and size can significantly influence how a document is perceived. Despite the availability of numerous font options, users often face challenges in selecting the most appropriate font for their specific needs. This research addresses this challenge by developing an automated font recommendation system that leverages machine learning to assist users in selecting optimal fonts for diverse document types.
A custom function was developed to automate the font download process, retrieving font files from the API and saving them for further analysis. In addition, another function was created to generate character images using the downloaded fonts, providing visual representations of how each font renders different characters. These images were crucial for training the DenseNet-201 model used in the recommendation system. By converting the fonts into 300x300 pixel grayscale images, the system was able to classify font styles and provide tailored font recommendations for various document types.
To build the system, a comprehensive dataset of font images was manually collected from Google Fonts using the Google Fonts API. The API allowed us to programmatically access font metadata and download URLs, facilitating the creation of a diverse dataset. Fonts were categorized into six main styles: Serif, Sans-serif, Script, Display, Monospaced, and Handwritten, representing a wide variety of fonts commonly used in document.
© 2025, IRJET
|
Impact Factor value: 8.315
This paper presents the process of dataset collection, image generation, and the subsequent application of machine
|
ISO 9001:2008 Certified Journal
|
Page 1063