DISA

When? Friday May 3 14-16
Where? Onsite: B1006 at Linnaeus University in Växjö and online
Registration: Please sign up for the PhD-seminar via this link https://forms.gle/9zwTEPoH9Uw26Cbi7 by May 1st (especially important if you plan on attending onsite so we have fika for everyone)

14.00-14.10 Welcome and practical information from Welf Löwe
14.10-14.55 Presentation and discussion: ChatGPT as a combined OCR and key-value extractor – Nemi Pelgrom, Industry PhD student at Fortnox
14.55 – 15.05 Coffee break
15.05 – 15.50 Presentation and discussion: Data-Driven Approach Towards Objective Evaluation of the Gravel Road Condition – Keegan Mbiyana, PhD-student in Mechanical Engineering
15.50 -16.00 Sum up and plan for our next seminar on May 31st

Abstracts

ChatGPT as a combined OCR and key-value extractor – Nemi Pelgrom, Industry PhD student at Fortnox

This study presents the results of an experiment of using the AI chatbot chatGPT as a combined OCR and key-value extractor. The dataset used consisted of 1000 real invoices and 500 images of real receipts from a wide variety of Swedish market sectors. This tool demonstrated an accuracy rate of 99.8% in extracting key financial data from invoices, and 100% accuracy from images of receipts, showcasing its potential as a one-stop solution for financial document processing.

Data-Driven Approach Towards Objective Evaluation of the Gravel Road Condition – Keegan Mbiyana, PhD-student in Mechanical Engineering

Visual windshield surveys are the primary method for assessing the condition of gravel roads, but they can be prone to human judgment errors and can be time-consuming. There is a need to shift to an objective assessment of the condition of gravel roads by leveraging advances in sensor technology and data processing techniques for efficiency and effectiveness. This seminar will present work conducted in the Department of Mechanical Engineering in objectively assessing the gravel road condition to improve maintenance planning and decision-making. The results from the field experiments on three gravel roads with different condition classes and the vehicle vibration response collected using Integrated Electronics Piezo-Electric (IEPE) and smartphone accelerometers to verify a proposed rule-based approach will be presented. The challenge of regular monitoring is discussed, considering the remote location of gravel roads. Regular condition monitoring and data collection would require collecting and storing huge amounts of data. Therefore, as part of future work, the seminar will briefly introduce an adaptive sampling technique to reduce redundant or unnecessary data before storage.