Wildfire Detection

An image classification pipeline that predicts whether a wildfire event occurs in the image.
Project Manager

Binghong Yu

Tech Lead

Yuxuan Fan

Developer

Shenghan Liu

Ruojia Tao

Wenbo Hu

Developer

Jerry Chan

Project Description

In recent years, wildfire has become a significant issue in California, causing enormous damage to the residents and the environment. To prevent the widespread of wildfires, applying machine learning algorithms to detect and forecast wildfires is a feasible path. This project will acquire video captures from surveillance cameras, detect the source of smoke and fire using machine learning algorithms, and send back the result to a wildfire detection network for further reactions and analysis.

Dataset

The dataset is collected by the High-Performance Wireless Research & Education Network (HPWREN) of UCSD and labeled by the SDSC researchers. HPWREN cameras are installed in remote locations to surveillance areas that are not well-covered by other technologies. The dataset contains screenshots from video snippets of wildfire events. The images are labeled positive or negative, depending on whether the screenshots contain fire or smoke.

Data Example

PositiveNegative
positive-datanegative-data

Get connected with us on social networks!

© 2021 — UCSD Triple C Light Mode