Detecting Parking Space with Generative AI and Large Multimodal Models

Abstract

RACHAEL POFFENBERGER, CHAZ CORNEJO, Dept of Computer Science, Mathematics and Engineering, Shepherd University, Shepherdstown, WV, 25443, and WEIDONG LIAO (Faculty Advisor), OSMAN GUZIDE (Faculty Advisor), Dept of Computer Science, Mathematics and Engineering, Shepherd University, Shepherdstown, WV, 25443. Detecting Parking Space with Generative AI and Large Multimodal Models

Generative AI and Large Multimodal Models have gained significant advancements in image recognition, detailed responses, and complex reasoning. In this project, we utilize these capabilities to monitor a parking lot and estimate the number of free parking spots with pictures captured with cameras placed above a parking lot. This system aims to provide real-time information to users, helping them determine if they should seek alternative parking. Additionally, with sufficient data and machine learning, the system can predict peak hours and busy days for the parking lot.

Despite the potential, current models sometimes struggle to produce accurate and consistent vehicle counts. This study focuses on improving the reliability and accuracy of the parking space monitoring system with comparative studies over various LMM models, architectural designs, and prompt engineering techniques.