Identificação automatizada de textos em imagens com Amazon Rekognition

• Autor(a) principal: Barata, Jardel Silas da Silva
• Data de Publicação: 2023
• Outros Autores: Teixeira, Lucas Ravele de Sousa, Silva, Bruno Campos da, Conte, Thiago Nicolau Magalhães de Souza, Santos, Wilker José Caminha dos
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Research, Society and Development
• Texto Completo: https://rsdjournal.org/index.php/rsd/article/view/40655
• topic: Amazon Rekognition; Deep learning; Text recognition; Data analysis.; Amazon Rekognition; Deep learning; Reconhecimento de texto; Análise de dados.; Amazon Rekognition; Aprendizaje profundo; Reconocimiento de texto; Análisis de datos.

The recognition of text in images is a challenge in the field of computer vision due to the variety of sources, image quality, size, and colors present in images. In this context, this work aims to develop an application for recognizing text in images using the Amazon Rekognition API and evaluate its accuracy. To achieve this, an algorithm based on deep learning techniques is proposed, capable of achieving an accuracy above 90% in the location and extraction of text in images, using data extraction methods from the text detection function of the Amazon Rekognition API. This article also has the potential to contribute to the advancement of future work in the field of computer vision, with a focus on text detection in images. Finally, the study concludes that the text detection API of Amazon Rekognition is relevant in data analysis, considering that it is trained with large amounts of image data to learn relevant characteristics, achieving an accuracy above 90%. However, it is necessary to consider that image quality and the type of font used can influence the accuracy of the results.