The risk of fires is ever increasing along with the boom of urban buildings. The current methods of detecting fire with the use of smoke sensors with large areas, however poses an issue. The introduction of video surveillance systems has given a great opportunity for identifying smoke and flame from faraway locations and tackles this risk. Processing this huge amount of data is a problem with using these video and image data. In recent times, a number of methods have been proposed to deal with this challenge and identify fire and smoke. Image processing algorithms for detecting flame and smoke, motion-based estimation of smoke, etc are some of the methods that are proposed earlier. Recently, there has been an array of methods proposed using Deep Learning, Convolutional Neural Networks (CNNs) to automatically detect and predict flame and smoke in videos and images. In this paper, we present a complete survey and analysis of these machine vision based fire/smoke detection methods and their performance. Firstly, we introduce the fundamentals of image processing methods, CNNs and their application prospect in video smoke and fire detection. Next, the existing datasets and summary of the recent methodologies used in this field are discussed. Finally, the challenges and suggested improvements to further the development of the application of CNNs in this field are discussed. CNNs are shown to have a great potential for smoke and fire detection and better development can help prepare a robust system that would greatly save human lives and monetary wealth from getting destroyed from fires. Finally, research guidelines are presented to fellow researchers regarding data augmentation, fire and smoke detection models which need to be investigated in the future to make progress in this crucial area of research.

中文翻译：

---

基于机器视觉的火灾探测技术：调查

随着城市建筑的蓬勃发展，火灾风险不断增加。然而，目前使用大面积烟雾传感器检测火灾的方法存在问题。视频监控系统的引入为识别远处的烟雾和火焰并解决这种风险提供了一个很好的机会。处理如此大量的数据是使用这些视频和图像数据的一个问题。最近，人们提出了许多方法来应对这一挑战并识别火灾和烟雾。用于检测火焰和烟雾的图像处理算法、基于运动的烟雾估计等是较早提出的一些方法。最近，已经提出了一系列使用深度学习的方法，卷积神经网络 (CNN) 可自动检测和预测视频和图像中的火焰和烟雾。在本文中，我们对这些基于机器视觉的火灾/烟雾检测方法及其性能进行了完整的调查和分析。首先，我们介绍了图像处理方法的基础知识、CNNs 及其在视频烟雾和火灾检测中的应用前景。接下来，讨论了该领域中使用的现有数据集和最新方法的摘要。最后，讨论了进一步发展 CNN 在该领域的应用所面临的挑战和建议的改进。CNN 被证明在烟雾和火灾探测方面具有巨大的潜力，更好的开发可以帮助准备一个强大的系统，该系统将大大挽救人类的生命和金钱，以免被火灾摧毁。最后，