Simone Smith
8th Grade, Saint Cyril of Alexandria School
Tucson, AZ
Catching Deadly Butterflies
Project Background
When Simone first learned about Artificial Intelligence Machine Learning (AIML) in 2022, she created programs to tell cats from dogs and chihuahuas from blueberry muffins. But once she had the hang of it, 鈥淚 wanted to apply what I learned about AIML from Girls Who Code to a real word problem to help people.鈥� She then learned about the use of PFM-1 mines, otherwise known as 鈥淏utterfly mines,鈥� being used in the war between Russia and Ukraine. The mines are the size of someone鈥檚 palm, bright green 鈥� and kids often mistake them for toys.
Tactics and Results
Simone decided to find out if an AI model could be trained to find PFM-1 landmines against different backgrounds. She first developed a 3-D printed model of the PFM-1, a small plastic butterfly shape that is designed to glide and scatter when dropped from the air. She spray painted her models green, brown or tan to mimic common colors. Simone took 100 photos of the mines against landscape features like rocks, grass, asphalt or dirt, and another 300 photos of the landscapes without the mines and fed the photos in to train her AI. She found out the AI model could spy the mines accurately about 90 percent against grass. Rocks and asphalt proved hardest as the varied rock background confused the model, and asphalt created shadows. But with more training, the model improved, and she ended up with 90 percent accuracy overall.
Beyond the Project
Simone鈥檚 favorite activity is swim team, 鈥渂ecause I am on a team with all types of people, learn to work hard, and have to do my best.鈥� She also likes to volunteer at the Tucson Village Farm as a junior camp counselor, where she plants and harvests vegetables, and helps campers gently handle the baby chickens. Simone would like to be a computer engineer. 鈥淚 especially like how computer engineers ensure that the hardware and software parts of the computers work together and do their job to accomplish a task.鈥�
