Azure Cognitive Services-Powered SaaS Invoice Recognition System
A leading software company offering a SAAS solution for invoice recognition to other businesses. The company wanted to provide a fully automated invoice recognition system to their clients to reduce manual efforts and improve accuracy.
The challenge was to create an accurate and efficient invoice recognition system that could handle a high volume of invoices across different vendors and formats. The system needed to recognize key invoice fields such as invoice number, date, issuer, line items, total and subtotal amount, tax values, etc. The Customer also wanted to provide the flexibility to train the system differently for specific clients, as well as allow manual adjustments in case of errors.
To meet the client’s requirements, Softellar’s team developed a cloud-native solution that used Azure Cognitive Services, RESTful API, Optical Character Recognition (OCR), and machine learning algorithms. .NET was used for the backend and React for the frontend. The solution was decomposed into microservices hosted in Service Fabric cluster and connected using Azure Service Bus, enabling scalability and flexibility.
The invoice recognition process involved uploading invoices to the system and training the system for specific vendors and invoice formats. The system used Azure Cognitive Services to extract key invoice fields, including line items, and verify the extracted data against the original invoice. In case of errors, the system allowed manual adjustments by the user.
To make recognition more accurate and allow manual adjustments OCR technologies were used, so it was possible to extract keys and values from the text manually when they’re not recognized automatically.
Result & Business Value
The solution enabled the Customer to provide a fully automated and accurate invoice recognition system to their clients. The system reduced manual efforts and improved accuracy, resulting in faster processing of invoices and improved customer satisfaction. The Customer also reported a significant reduction in errors and manual corrections, leading to cost savings and increased efficiency.
Tools & Technologies
.NET, C#, React, TypeScript, Azure Service Fabric, Azure Service Bus, Azure Cognitive Services, OCR, Machine Learning