Faculty Supervisor
Dr. Ghassem Tofighi
Date of Defense
Fall 12-2020
Program Name
Honours Bachelor of Computer Science (Mobile Computing)
School
Applied Computing
Keywords
object detection, fire code violation, fire code violation detection, depth detection, machine learning, YOLO
Department
Faculty of Applied Science & Technology (FAST)
Description
This paper showcases the use of a machine learning algorithm in order to detect fire code violations.
Abstract
his paper explores the creation of an object detection system for mobile using YOLO(You Only Look Once) algorithm., a real-time object detection model that is developed to run on a portable device such as a cellphone that does not have a Graphics Processing Unit (GPU). This algorithm is utilized to detect fire code violations, specifically the obstructed door in a fire separation: the areas surround- ing the door opening shall be kept clear of anything that would be likely to ob- struct. The machine learning algorithm utilized has been fine-tuned to fit the model based on accuracy levels. The author has run multiple experiments to determine the best accuracy levels, which highlights the importance of hyperparameter opti- mizations in improving the performance of an object detection model. The results of the experiments also have shown that training beyond 5000 iterations can produce an overfitting model. Furthermore, the author discusses the combination of depth detection with object detection, to increase the accuracy of the violation detection.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Elewa, Salim, "Fire Code Violation Detection" (2020). Student Theses. 7.
https://source.sheridancollege.ca/fast_sw_mobile_computing_theses/7