PROJECT TITLE :

Optimal Placement and Intelligent Smoke Detection Algorithm for Wildfire-Monitoring Cameras

ABSTRACT:

Wildfire smoke is frequently evident much before flames are visible. To avoid major property losses and heavy mortality from catastrophic wildfires, early identification of wildfire smoke is critical. To identify wildfire smoke in a timely manner, camera networks are being developed and extended. An intelligent video smoke detection algorithm and an optimal wildfire camera placement plan are crucial for achieving the highest camera coverage and detection accuracy with a restricted budget. We present an effective video smoke detection system for embedded applications on local cameras in this research. It is made up of two modules. Local binary patterns and a dense optical flow estimator are used to process the original video frames in the first module. The created features are then sent into a lightweight deep Convolutional Neural Network, which acts as a binary classifier to detect the presence of smoke in the second module. To reduce the overall fire danger of a given area, we also formulate the wildfire camera placement problem as a binary integer programming problem. Case studies using real-world movies are used to test the proposed smoke detection framework's accuracy, as well as its computational and memory efficiency. By replicating the deployment of wildfire cameras across a test region in Southern California, we also confirm our suggested camera placement approach.