A-Maize-ing Drones

Inspiration

Our team was inspired by the thought of helping hardworking farmers whilst also providing solutions to the UN's 2nd SDG. We started by thinking about where we could add the most value using drones, AI, Edge Computing and

What it does

Our prototype solution uses image recognition and machine learning to detect weeds in fields of crops. Edge computing enables us to use the power of machine learning in remote areas. This technology allows farmers to save time when farming by being alerted to the exact position of weeds. It also allows farmers to save money by reducing the use of herbicides, which are generally harmful to the environment they're used in.

How we built it

We started out by finding a dataset online, which contained labeled images of soil, grass, soybeans and Broadleaf weeds. We then ran the images through a machine learning algorithm that trained an image recognition model which is to be used by the drone via Edge Computing.

Challenges we ran into

Finding the ideal data set was one of our first challenges and it largely dictated the prototype target environment (i.e soybean crops instead of maize crops - which we initially wanted to go for). Size and the quality of data were key concerns when we were trying to find an adequate data set. We ran into significant technical challenges with trying to engage with the drone simulator. This meant that we were unable to devote sufficient time to test the AI on live data.

Accomplishments that we're proud of

We are very proud of the way we took onboard new challenges, such as dealing with drone software for the first time. One of our members had his first exposure to neural networks and he is very proud of himself for the work he has completed so far.

What we learned

We learned a lot about the use of emerging technologies to solve some of the world's most pertinent problems. On the logistical side of things, we were able to learn about how to work strategically for 24 consecutive hours and play to each other's strengths. Teamwork makes the dream work!

What's next for A-Maize-ing Drones

We look forward to implementing recognition for more species of weeds as well as using live image fees in our AI model. In the future, we hope our drone technology can receive more inputs such as humidity and temperature to allow for precise monitoring of crop growth conditions. With more work, we also hope that we will be able to determine the peak ripeness of crops so they can be harvested in good time.