Aplicações de técnicas de reconhecimento óptico de caracteres para a extração de texto em imagens: um estudo de caso no comércio varejista de artigos de vestuário

Abstract

The objective of this study was to develop a computational tool for the retail sector, aiming to optimize the management of supplier data through the automation of information extraction from business cards. The tool uses Optical Character Recognition (OCR) techniques, Natural Language Processing (NLP) with Named Entity Recognition (NER), and regular expressions to identify data such as phone numbers and Instagram profiles. The process was divided into data collection, model testing, validation, and the implementation of a graphical interface to organize the extracted data. The database consisted of 21 business card images. After validation, the tool demonstrated good performance in extracting and organizing the information, with satisfactory results for standardized data, although expanding the database is necessary to improve the model's robustness. This work contributes to the digitalization and automation of supplier management processes in the retail sector, opening opportunities for future improvements and research in the field.