Avinash Sri-Kumeran , NATURE Lover, ASPIRING Wildfire Early-Detector

The 85,000-acre CZU complex wildfire was pivotal in my early high school years. My Godfather lost his mobility, which resulted in me losing my bicycle companion, which catapulted me into searching for solutions to predict wildfires (Aug 2020)

Early Prediction of Wildfire using Machine learning Algorithms

I’ve been deeply intrigued by the prospect of forecasting wildfires in advance, which motivated me to complete research on the topic successfully.

CZU lightning complex fire

I realized that my best consultants in this undertaking would be the Fire Department Captains of nearby stations. Consequently, I commenced a routine of visiting them practically every weekend. They proved to be more than willing to share their knowledge with me, elaborating on the intricacies of fire trucks and equipment. Some even proudly showcased their latest trucks and cutting-edge technologies. Over time, a genuine friendship blossomed among us. The more time I spent with them, the more insights I gained. I took food to them during the holidays and enjoyed celebrating Independence Day at the fire station. I discerned that three fundamental aspects held paramount importance to them:

Achieving absolute accuracy in detecting fires
Receiving early notifications of fire outbreaks
Effectively categorizing the nature of fires

(Fall 2020)

Fire Marshall Knight, Santa Clara Fire Station
Fire Captain Mark Franz, West Valley Fire Station
Several other fire fire fighters became my advisors in this journey.

I immersed myself in research papers roughly a year into my quest for a solution. I conversed with professionals from CalFire and the San Jose State University WIRC team. This comprehensive understanding prompted me to design my drone with infrared sensors and LIDAR technology. This endeavor was challenging, as I underwent a series of trial and error phases. Nearly an entire summer was dedicated to the construction of this drone. (Summer 2021)

( Summer 2022) : Subsequently, I came to realize that a solid grasp of machine learning algorithms was imperative for predicting fires. During the December holidays of 2022, I dedicated myself to mastering these algorithms. As the spring of 2023 unfolded, I was deeply engrossed in coding machine learning algorithms to forecast fires. This marked a turning point in my journey. To refine my ideas, I sought counsel from individuals

Mentor: Juyon Lee
- B.SC – Computer Science Stanford University,
- M.SC – Computer Science Oxford University
Prof. Ali Saeidi, DeAnza College
Kurtis Nelson, Earth Resources Observation and Science (EROS) Center

63/538,870 (patent pending)

avinash-sri-kumeran-early-prediction-of-wildfires-paper Download

( Summer2023 ) : Today, I am ecstatic to announce the culmination of my efforts: a deep learning algorithm capable of predicting fires with an accuracy exceeding 90%. This is undoubtedly a defining moment for me. I am publishing my research paper ( download link provided above) this fall in several science scholastic magazines.

( Fall 2022) In order bring my peers, friends and family to the cause, I have created a non-profit org – CaptivatingSpark.

Captivating Spark – 501c3 Corporation:

My personal commitment remains resolute in leading CaptivatingSpark towards its dual-fold objectives. Our mission encompasses help with early notification while concurrently extending assistance to fire victims both within the United States and globally.