Implementation of an enhanced parking occupancy detection system using Convolutional Neural Networks

Responsable: Leyre Encío [leyre.encio@upm.es]

 

This master’s thesis introduces an improved parking occupancy detection system designed to accurately identify parking slot occupancy even in challenging, barely visible scenarios. Employing deep learning techniques with Keras and TensorFlow, the system employs Convolutional Neural Networks (CNNs) and image processing to achieve high accuracy. The objective of this master thesis is to detail the model’s architecture and training process and demonstrate its real-time performance in a practical parking environment. A video demo of the system output is available at https://youtu.be/yuhue6VVhlM. This contribution aims to enhance urban mobility, reduce congestion, and optimize parking space utilization.