TEXT RECOGNITION USING COGNITIVE SERVICES IN AMAZON WEB SERVICES
DOI:
https://doi.org/10.24867/24BE14KnezevicKeywords:
Amazon Web Services, cognitive services, machine learning, Textract, Lambda functionAbstract
This paper is based on an exploration of the use of machine learning for recognition and extraction of text and data from sets of documents in different formats, which contain images of texts written by hand or in different languages, printed letters in digital format. The research is focused on the use of Amazon Textract, a cognitive service provided by Amazon Web Services, for automatic document processing. The goal of the research is to use different inputs to collect results that will be used to test and analyze the accuracy of text recognition and document processing performance by the Amazon Textract service, as well as adaptability to different input documents and the quality of input documents. The following Amazon services were used for the technical implementation of the task: Simple Storage Service. Lambda function and Textract service.
References
[2] Artificial intelligence services AWS, preuzeto sa https://aws.amazon.com/machine-learning/ai-services/
[3] Amazon Textract, preuzeto sa https://aws.amazon.com/textract/
[4] AWS Lambda Function, preuzeto sa: https://docs.aws.amazon.com/lambda/latest/dg/welcome.html
[5] Amazon Simple Storage Service, preuzeto sa https://aws.amazon.com/s3/
[6] Python, preuzeto sa https://www.python.org/
[7] Suminda Niroshan, “AWS Textract with Lambda Walkthrough” (28. Jun 2019), preuzeto sa: https://medium.com/@sumindaniro/aws-textract-with-lambda-walkthrough-ed4473aedd9d
[8] Dr Urs Marti, Institut za računarske nauke i poslovnu matematiku, ETH Zurich, Švajcarska, “IAM Handwriting Database”, (1999) preuzeto sa: https://fki.tic.heia-fr.ch/databases/iam-handwriting-database
[9] Nikolaos Arvanitopoulos, Gaspard Chevassus, Daniele Maggetti, Sabine Süsstrunk, “A Handwritten French Dataset for Word Spotting: CFRAMUZ” (Novembar 2017), preuzeto sa https://dl.acm.org/doi/10.1145/3151509.3151523
[10] Jens Walter, “my receipts (pdf scans)”, preuzeto sa: https://www.kaggle.com/datasets/jenswalter/receipts
[11] Francisco Cruz, Mauro Castelli “Dataset of invoices and receipts including annotation of relevant fields” (21. Mart 2022), preuzeto sa https://zenodo.org/record/6371710#.ZAHGW3aZOUl
