Detecting Human Life During Fire

ABSTRACT

This project focuses on the development and evaluation of a robust detection system for identifying humans, fire, and smoke in real-time using advanced deep learning algorithms YOLOv8 and YOLOv9. Leveraging a dataset from Roboflow that contains a variety of images featuring these elements, we aim to implement a model that accurately detects and classifies the presence of humans, fire, and smoke in uploaded images and live camera feeds.The system will be developed in Python using Google Colab as the integrated development environment. By employing the capabilities of YOLOv8 and YOLOv9, we will compare their performance in terms of accuracy, speed, and robustness in various detection scenarios. The model will be capable of processing images from a laptop camera, although we acknowledge potential limitations due to the camera's resolution and clarity, which may affect detection accuracy.The expected outcome is a functional application that provides real-time alerts and visual feedback when fire, smoke, or human presence is detected, enhancing safety and response measures in critical situations. This project aims to contribute to the growing field of computer vision and its applications in safety and security systems.

Keywords: Real-time detection system, humans, fire, smoke, YOLOv8, YOLOv9, deep learning, Roboflow dataset, image classification, live camera feeds, Python, Google Colab, accuracy, speed, safety, security, computer vision.