About Me
- Name: Mansur Yassin
- School: University of Washington, Computer Science and Systems
- Email: mansur.yassin00@proton.me
About My Mentor
- Name: Jeremy Waisome
- School: University of Florida, Department of Engineering Education
- Area of Research: Dr. Jeremy Waisome’s research interests lie in broadening participation in STEM, particularly for historically underrepresented groups. She focuses on creating supportive educational environments that encourage the persistence and success of minority students in STEM disciplines.
- Link to Their Website: Jeremy Waisome
About My Project
The NSF research project focuses on understanding plant roots and their adaptation to changing climates by using advanced sensor technologies like minirhizotron (MR) systems. These systems capture color images of roots through cameras placed in the soil, but preparing these images for scientific research is labor-intensive and requires human interpretation. The project aims to develop machine learning tools to automate this process and improve the collaboration between humans and machine learning algorithms. By involving plant scientists of varying expertise, including novices and youth, the project seeks to enhance the interactive machine learning experience and encourage more students to see themselves as future scientists. This effort includes participatory co-design workshops and summer science experiences with Florida 4-H programs. The outcomes will advance both the data analysis of MR systems in plant science and the broader application of human-centered machine learning in other fields like human anatomy and hydrology. Ultimately, the project aims to increase productivity, sustainability, and resilience in agricultural and natural ecosystems while also expanding the STEM workforce by involving marginalized youth. More details can be found here.
This project presented by the Social computing for good lab, developes and enhances the accuracy of election ballot verification systems through the use of Informed Optical Character Recognition (iOCR). This project addresses the limitations of traditional Optical Character Recognition (OCR) systems, particularly in the context of structured documents like election ballots, where formatting inconsistencies and variations in text can result in recognition errors. In election systems, these errors can compromise the integrity of vote tallies, making accurate interpretation of voter intent crucial. One of the motivations behind this research is to offer an alternative to the existing Ballot Marking Devices (BMDs) that encode votes in barcodes. These barcodes are not human readable by raising transparency concerns since voters cannot directly verify the encoded information. The research aims to replace barcodes with OCR systems that interpret human readable text which increases transparency and trust in the election process. However, traditional OCR systems often introduce recognition errors that may lead to discrepancies between the voter’s intent and the recorded vote. To mitigate these issues, the research explores the use of the Levenshtein distance algorithm to correct OCR errors. This error correction technique compares the OCR output with the original ballot definition by calculating the number of character edits required to match the scanned text with the expected entries. When integrating this post processing technique, the iOCR system enhances the accuracy of ballot verification and the scanned results should align with the voter’s intended selections. The purpose of this research is to develop a more accurate, transparent, and reliable method for verifying election ballots, ultimately offering a solution that combines the accessibility of human-readable text with realiable machine verification. The iOCR system is designed to correct errors introduced by traditional OCR methods and provide an improved framework for secure and transparent elections. here
My Blog
Week 1, Week 2, Week 3, Week 4, Week 5, Week 6, Week 7, Week 8, Week 9,