Overview

This project automates receipt processing using AWS Textract and Step Functions, with a focus on comparing different machine learning models for data extraction. The repository includes Jupyter notebooks demonstrating:

• Uploading receipts and processing them using AWS Textract
• Implementing AWS Step Functions to orchestrate the workflow
• Comparing multiple machine learning models for improved accuracy

Repository Contents

1. UploadReceiptsSaveTextract.ipynb

This notebook handles:

• Uploading receipt images to an S3 bucket
• Extracting text from receipts using AWS Textract
• Saving the extracted data in a structured format for further processing

2. ModelComparison.ipynb

This notebook focuses on:

• Comparing different machine learning models for receipt data classification
• Evaluating model performance using accuracy metrics
• Visualizing model results

3. StepFunctionDescription.ipynb

This notebook explains:

• The design and implementation of AWS Step Functions as a pipeline to process data
• Users on the Android app interact with the Step function using a REST API
• The state transitions involved in processing receipts
• How extracted data is stored and analyzed

Key Highlights

• Integration of AWS Textract for automated text extraction from receipts
• Use of AWS Step Functions to manage and orchestrate workflows
• Application of machine learning techniques to classify and analyze extracted data
• Python-driven implementation with structured, reusable Jupyter Notebook workflows
• Data visualization techniques to compare and evaluate model performance