This report details the development of a software program aimed at converting handwritten parachute data records from the Natick Army Lab into an analyzable digital format. Due to the failure of traditional OCR models to recognize handwritten fractions, a crucial part of the data, we generated a synthetic dataset to specifically fine-tune these models. The software utilizes image processing and OCR technologies to translate text and replicate the physical document's layout in a digital format. This innovation streamlines data analysis, enhancing the Army's ability to monitor and understand parachute integrity and lifecycle.

• This report represents the work of one or more WPI undergraduate students submitted to the faculty as evidence of completion of a degree requirement. WPI routinely publishes these reports on its website without editorial or peer review.

Creator

• Norton, Cameron
• Langevin, Christopher
• Hall, Liam
• Patel, Harsh
• Haley, Matthew
• Trilling, Elliot

Subject

• Innovation
• Industry
• Project-Based Learning
• Computing

Publisher

• Worcester Polytechnic Institute

Identifier

• E-project-041724-161231
• 121155

Keyword

• Transfer Learning
• Machine Learning
• Optical Character Recognition
• Image Processing

Advisor

• Paffenroth, Randy Clinton
• Mangoubi, Oren Rami

Year

• 2024

Sponsor

• Professor Gregory Noetscher
• U.S. Army Natick Soldier Systems Center

UN Sustainable Development Goals

• 9 - Industry, Innovation and Infrastructure

Date created

• 2024-04-17

Resource type

• Major Qualifying Project

Major

• Computer Science
• Mathematical Sciences
• Data Science

Source

• E-project-041724-161231

Rights statement

• In Copyright