Forest Fire Detection Network using Wireless Video and Sensor Identification

Forest fires are naturally occurring disasters that have the potential to be very destructive if not detected and controlled in time. Several papers have been published mentioning the use of sensors or cameras to detect fires; however, a small minority have combined both technologies to obtain better results in detection. 

Our project proposes a solution to forest fire detection through the usage of automated nodes equipped with cameras and sensors. The nodes are comprised of an ESP-8266 Microcontroller to which a DHT-11 Temperature and Humidity Sensor, as well as an MQ-2 Gas Sensor, are connected, and a Raspberry Pi 4 running a web server that can receive pictures from different sources to detect fires through a YOLOv8 image recognition model.

The sensor readings can be observed on a web application that was deployed on AWS. The frontend displays the different forests with the nodes present in each one, along with the danger level of each node which is dynamically updated by the backend according to sensor readings and camera detections.capstone_ece_24-25_ForestFireDetection_pic3.jpg

 

Project Details

[photo]