What it does
Light Weight Fire Detector. Using a deep learning neural network model to train on a set of images and do predictions on images of Fire. The trained model is launched as a servable API using flask and rock for port-forwarding from local machine to google colab.
How we built it
Using Neural Network Deep learning, Convolution Model, Google Colab, Flask, Ngrok, Web Dev. Keras, TensorFlow
Challenges we ran into
Training the model to work well on new test data. Using regularization techniques to solve overfitting. Exposing the trained model to
Accomplishments that we're proud of
Learned to apply deep learning and explored new ways to make an API public without hosting it on cloud.
What we learned
Neural Network, Web Dev, How to fail and rebuild
What's next for Light Weight Fire Detector
To identify ways to for data sources through cheap and easily accessible hardware in home safety.