Earlier this week, Peachtree Ridge High School senior Shaza Mehdi presented during the keynote session at Google I/O 2018, the tech giant’s annual developers conference in Mountain View, California. There, Shaza demoed PlantMD — a smartphone app she developed that helps identify different types of plants and plant diseases.
 
“My mom grows rose bushes in our front yard, and every season they would get diseased,” Shaza said. “I wanted to have a way people could diagnose plant diseases just by taking a photo of it.”
 
“Knowing Shaza, she wanted to do something about it,” Shaza’s father Salman Mehdi said.
 
So, she took to video tutorials and blog posts to learn how it could be done. That’s when she discovered Google’s open source machine learning framework TensorFlow.
 
Using TensorFlow’s artificial intelligence computer vision capabilities, PlantMD identifies plants and plant diseases by simply analyzing a photo of a plant.
 
The app converts each pixel of an image into an RGB value and then compares the RGB values to a library of other plant images to identify similarities and detect patterns.
 
Over time, as more users upload photos to the dataset, the app becomes more and more accurate.
 
Shaza presented her idea and progress to developers from around the globe at Google I/O. The app is available for download on Android devices.
 
What will Shaza do next? She’s now working on replicating the success of PlantMD and using similar machine learning and convolutional neural networks to develop an app that identifies common human skin conditions.