Image analytics using artificial intelligence (fire and smoke detection)

Abstract

With the cost of human labor on-site increasing annually, faster and more accurate technology for detecting fire without constant supervision is necessary. Specifically, this report will use a YOLO (you only look once) model powered by Artificial Intelligence (AI). Currently, YOLO has 11 versions, and we will be using YOLO V10 as the base model in this project. This system uses convolutional neural networks (CNNS) to process and analyze real-time video feeds from surveillance cameras. The proposed method offers several advantages, including detecting fires at their earliest stages. By training the AI model on a comprehensive dataset containing diverse fire and smoke scenarios, the system learns to identify patterns and characteristics of fire and smoke. The system will detect fire or smoke without the presence of conventional smoke detectors, which can only be installed in enclosed environments. Although false positive results may be present, the threat of fire is too significant to ignore for anyone's safety. This will enhance fire safety measures and solutions for users, paving the way for more intelligent and more efficient fire detection and prevention approaches.