datefind

Inspiration

The inspiration for this project came from the need to simplify financial management. The more bills and due dates one has to keep track of, the more it becomes overwhelming with other things to attend to. I wanted a tool that would automatically do this process for me and reduce the chances of missing payments.

What I Learned

The project taught me useful skills in: - Optical Character Recognition (OCR): Extracting text from images using OCR technology. - API Integration: Interacting with Google Calendar API to schedule events programmatically. - Automation: Designing workflows to reduce manual effort in repetitive tasks. - Error Handling: Managing edge cases, such as illegible text or unsupported file formats.

Also, I learned how to integrate machine learning models and APIs into real-world applications.

How I Built It

1. Scanning Bills:

   • Implemented OCR using Python's pytesseract library for text extraction from the scanned bills.
   • Processed the extracted text to extract key information like the bill amount, due date, and vendor name.

2. Parsing Information:

   • Used regular expressions to extract structured data, such as dates and payment amounts, from the raw OCR output.

3. Integration with Google Calendar:

   • Utilized the Google Calendar API to create events for each due date of a bill.
   • Automated adding reminders to alert users before the due date.

4. User Interface:

   • Developed a basic command-line interface (CLI) for initial interaction.
   • Enhanced the interface to be desktop or mobile-friendly, which will be implemented in future versions.

Challenges Faced

1. Accuracy of OCR:

   • The greatest challenge was to extract the accurate information from the bills with diversified layouts, fonts, and quality. • Improved OCR accuracy by pre-processing images: rescaling, sharpening, and noise removal.

2. Data Extraction:

   • Date and amount parsing always required adjusting regular expressions for each new format.

3. API Integration:

   • Authentication and permissions with Google Calendar API. There were a few bumps in the road to start with.

4. Error Handling: • Handling of unreadable scans or unexpected input formats in a graceful way.

Conclusion

The project made the management of bills much easier and efficient, but most importantly, it gave impetus to the importance of automation in daily life. Going forward, I would like to continue improving the application with added support for recurring bills and a friendlier interface.

Built With

• python