Inspiration
When I was still living in Zambia(Africa), my family and I would explore different rainforests all over the world and those are some of my best memories; I can't imagine going back to visit, and not being able to visit my favorite forests, and deserts so I knew I had to do something about it.
What it does
ThermalAI uses Thermal Vision, Computer Vision and Scientific computation to help ensure our biomes are staying at healthy temperatures. For example our Ice caps are melting at exponential rates due to global warming; What if we had a model that could actively measure these temperatures and identify the specific areas that need some extra attention(Maybe it would also lessen response time to certain disasters such as forest fires)? That is exactly what ThermalAI does! We use computer vision in real time with thermal vision to measure the temperatures of various environments and all of our data is stored in Google Cloud(As data scientists it's important to keep data in order to identify trends and make smart decisions).
How we built it
This application was built using openCV for computer vision, Yolov5 for real time object detection, Numpy for scientific computation, and Google Cloud for data storage.
Challenges we ran into
I truly struggled with finding training data that was large enough to train my mode, but also not too large so I can train in within a few hours. Realistically using a laptop camera for thermal vision is very limiting, I wish I had some hardware to be able to better simulate this. I would have liked to make my model be able to detect fires or different weather conditions that affect our biomes, but that proved to be difficult because it's hard to demonstrate/simulate that in real time.
Accomplishments that we're proud of
I switched projects halfway through the competition, but I never gave up. I had a very hard time creating a structure for my project, and trusting the process. I am very new to the data world, but I love it and this challenge truly helped me grow my confidence as a developer. I really walked “into the unknown”, but I feel very successful.
What we learned
I learned to trust myself, and to take a step back when things get hard. As developers we are really close to our projects, which directly makes us miss very small issues! For example, I spent 2-5AM trying to implement a feature for my project, after my power nap I was able to fix my issue in 7 minutes.
What's next for ThermalAI
I would like to implement a camera that is better suited for Thermal Vision for better accuracy. I think I would also love to explore more datasets and improve my confidence in Machine Learning algorithms so I can maximize the accuracy of my application.
Log in or sign up for Devpost to join the conversation.