Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
Eine Plattform für die Wissenschaft: Bauingenieurwesen, Architektur und Urbanistik
An Emerging Fire Detection System based on Convolutional Neural Network and Aerial-Based Forest Fire Identification
Accidents caused by undiscovered fires have resulted in substantial financial losses around the world, emphasizing the critical need for robust fire detection systems. Existing fire and smoke detectors have demonstrated inefficiencies, forcing the development of more sophisticated alternatives. Based on Convolutional Neural Network (CNN) technology, this research provides a novel fire detection system. By researching fire flame features and utilising edge detection and thresholding algorithms, the system accomplishes real-time fire detection by analyzing live camera data. A fire detection model is developed that can detect dangerous fires based on their size, velocity, volume, and texture. The experimental findings of our model on a fire dataset that we created ourselves show its strong fire detection capability, especially in real-time multi-scale fire detection. We use the YOLOv4 protocol, the Darknet deep learning framework, and a new network topology to build the fire detection model. A large-scale, multi-scale system is used for extensive training and testing, outperforming previous methods in terms of the accuracy of fire detection.
An Emerging Fire Detection System based on Convolutional Neural Network and Aerial-Based Forest Fire Identification
Accidents caused by undiscovered fires have resulted in substantial financial losses around the world, emphasizing the critical need for robust fire detection systems. Existing fire and smoke detectors have demonstrated inefficiencies, forcing the development of more sophisticated alternatives. Based on Convolutional Neural Network (CNN) technology, this research provides a novel fire detection system. By researching fire flame features and utilising edge detection and thresholding algorithms, the system accomplishes real-time fire detection by analyzing live camera data. A fire detection model is developed that can detect dangerous fires based on their size, velocity, volume, and texture. The experimental findings of our model on a fire dataset that we created ourselves show its strong fire detection capability, especially in real-time multi-scale fire detection. We use the YOLOv4 protocol, the Darknet deep learning framework, and a new network topology to build the fire detection model. A large-scale, multi-scale system is used for extensive training and testing, outperforming previous methods in terms of the accuracy of fire detection.
An Emerging Fire Detection System based on Convolutional Neural Network and Aerial-Based Forest Fire Identification
Goel, Akanksha (Autor:in) / Bhende, Manisha (Autor:in) / Lal, Mily (Autor:in) / Pathak, Abha (Autor:in) / Tamrakar, Poi (Autor:in) / Thorat, Pallavi (Autor:in) / Sharma, Swati (Autor:in)
10.12.2023
715030 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Drone Based Fire Detection System Based on Convolutional Neural Network
| BASE | 2024
Hangar and forest fire monitoring system based on unmanned aerial vehicle
| Europäisches Patentamt | 2021