Week 9
As I spent my last two weeks completing my final report, my main goal was to provide a clear and detailed account of the entire development process; While drafting the report, I focused on explaining the significance of implementing regex based text cleaning to handle the noisy OCR output. It was important to highlight how this step made a crucial difference in ensuring that unnecessary characters and formatting inconsistencies were removed. I also made sure to clearly describe how the regex pattern worked in the system and how it allowed us to extract only the relevant ballot data for further processing. Documenting the application of the Levenshtein distance algorithm was another key part of the report. I wanted to ensure that readers understood how the algorithm was used to correct minor OCR errors, such as character misspellings and formatting variations. Writing this section required careful explanation of how the algorithm helped improve the system’s accuracy in recognizing candidate names and propositions, which was essential to the success of the project. As I wrote the results section, I reflected on the experiments we conducted to compare the iOCR system against traditional OCR systems. The key was to convey the system’s improvement, particularly how switching to the 10 line structure and implementing regex validation helped us achieve 100% accuracy. Detailing the edge cases we encountered, such as lines that were too short or too long, was critical in showing how the iOCR system would handle real world challenges effectively. Lastly, in the report’s conclusion, I emphasized the significance of the iOCR system for improving election ballot verification processes. I aimed to communicate how this project not only improved OCR accuracy but also offered a more transparent and reliable method of ballot processing. Writing the report provided an opportunity to reflect on the technical achievements and the lessons learned throughout the project. It also allowed me to document the process in a way that could be useful for future improvements and similar applications of OCR in other fields.