Development of Fire Detection Using Machine Learning /
John Kyle R. Almirol, Roxanne Janella R. Hicban, Veren Dale L. Leoncio, Dana Trisha B. Valencia.
- x, 107 pages : illustrations ; 29 cm. + 1 CD-ROM (4 3/4 in.)
Thesis (Undergraduate)
College of Science --
Includes bibliographical references.
Most fires are caused by runaway reactions, operation error and equipment failure, and flammable material release. These mostly occurred in working places of no prompt fire protection. The Development of Fire Detection using Machine Learning as a key component of fire protection and safety plan in chemical laboratories, aimed to detect the unwanted presence of fire in order to provide responses to a deemed threat to people, property, and environment. The concept of this project is to minimize the risk and maximize the safety for chemical laboratories. Fire detection includes a camera, gas sensor and smoke sensor connected to the Raspberry pi 4 and Arduino. In addition, an email notification will be added to the system so that the user is instantly notified, resulting in early detection. Numerous lives can be saved by early fire detection, which can also prevent irreversible infrastructure damage to properties and ensuring the financial losses. A possibly fatal incident with great damage is fire. A fire alarm is a vitally important part of any structure. It must be able to provide accurate and useful detection. Using the currently available techniques of smoke sensors installed in the structures, fire detection can be very challenging. Due to their outdated technology and design, they are costly and slow. The possible use of artificial intelligence for detection and sending alerts with video from camera footages and sensor detection based was critically examined in this research. The smoke sensor and gas sensor which are utilized as a part of the Fire Detection System to recognize a fire. The smoke sensor detects the present particles in the room. The gas sensor detects if there is any gas present in the room. To officially execute all of the commands from the two sensors, Arduino Uno was used. A self-built dataset of video frames with fire was also used. One of OpenC V's best features is the wide range of built-in primitives it offers for handling tasks related to image processing and computer vision. It is possible to achieve better outcomes than traditional systems with the use of sensors combined with computer vision and image processing techniques. The developed fire detection system was subjected to evaluation and results revealed that it is with this the proposed system can provide more reliable information.--Author's Abstract.