Submission: HackeRUs

[samwilson0745]

Team Name

HackeRUs

Team members

Member # 1
Name: Soham Deb Majumder
Email: [email protected]
Twitter handle (hyperlinked): https://twitter.com/SohamDebMajumd1

Member # 2
Name: Arkadeep Mitra
Twitter handle (hyperlinked):

Member # 3
Name: Deepshika Biswas
Twitter handle (hyperlinked):

Project Name

Real-time Fire Detection System using Machine Learning

Contact Details (Leader)

[email protected]

Project Track

Community

Link to project GitHub public repo

https://github.com/samwilson0745/Fire_Detection

Link to project website

-n/a-

Link to demo video

https://youtu.be/MqnoTtRWJ_c

Inspiration

Inspiration

The inspiration for this project came from the horrific experiences of the locals of Australia who experienced first hand the destructive fury of Mother Nature who, claimed countless lives in the Great Australian Forest Fire and thus a need arose of a surveillance system to overview the forest so that such tragedies can be immediately reported and necessary precautions can be taken beforehand.

What it does

Fire Detection Model was made using Machine Learning with Python models such as Tensorflow, numpy, openCV, ImageDetection among other.

Over 100 files and 5000 steps to maximize accuracy.

The model can detect Major Fires upto 60% accuracy and Minor Fires upto 55% accuracy. Non Fire situations can be detected with 72% accuracy.

Challenges you ran into

• First challenge was the lack of datasets containing pictorial information of fires.
• Secondly we faced was the lack of trainable machines and thus we trained our own machine to classify fires.
• Using google colab as a web based platform for our jupyter notebook which lacked the necessary RAM for our needs.

Anything else?

No response