Large-scale Fire Detection based on YOLOv8 Lightweight Model and its Improved System

In recent years, global warming, population increase, the expansion of human production and living areas and many other factors have led to the increasing frequency of forest fires, forest fire prevention and response has become more and more important. Nowadays, computer vision methods and video surveillance cameras have a wide range of applications, has shifted from traditional detection methods to machine deep learning technology, these technologies have a very important role in promoting and significance of fire detection, image detection can be extracted through the dataset training after the object features, compared with the traditional fire detection methods, it can detect and warn of fires in a short period of time, and the coverage of the area is wider, and can realize a wide range of detection. It can realize a wide range of detection. YOLOv8 and its improved models have great advantages in this regard. The application of YOLOv8-based model in fire detection can greatly reduce the consumption of manpower and financial resources for detection, and reduce the economic losses caused by fire. This article mainly tested yolov8 with Fire smoke detection and compared it with convolutional neural network and multimodal model as well as based on yolov8 model by adding modification module to achieve light weight, wide range detection and other functions.