Inspiration

Algae blooms are dangerous environmental events with potentially catastrophic consequences. Algae Vision is designed to be able to detect these events and warn users before they get out of hand.

What it does

Algae Vision uses two convolutional neural networks to take an image of terrain, find the area with water, and detect the amount of algae present in the water.

How we built it

We used two convolutional neural networks in TensorFlow to first take an image of terrain and detect where the water is in the image, then to use characteristics of the water to find a percentage probability that dangerous algae is present.

What we learned

When we came into this hackathon, we had no idea how to train a AI - we spent a lot of time teaching ourselves how to use TensorFlow.

What's next for Algae Vision

Algae Vision's technology could be deployed on specialised satellites which constantly fly over and scan the terrain below and watch for these blooms, sending an alert to an environmental station whenever there might be a problem. Monitoring algae blooms is crucial because these blooms can absolutely devastate ecosystems, so being able to detect these before they get out of hand will greatly help save the environment and counter the effects of climate change.

Built With

Share this project:

Updates