There are 5.7 million Americans living with Alzheimer's disease. The caring involuntary help has been valued over 232 billion USD. When one of the founders of ALZlens came to the US, he interacted with an Alzheimer's patient and was deeply compelled by the challenge of "no technology" in the sector.
What it does
AlzLens uses Amazon's Rekognition software to compare a taken photo with a pre-uploaded selection of faces of different people. If a match is detected, AlzLens displays the name of the person on an LCD screen.
How I built it
AlzLens uses a Raspberry Pi connected to a breadboard with a pull-down resistor button and a breadboard with a 16x2 LCD screen. We used an open-source software found on GitHub as the building blocks used to upload reference pictures to Amazon Rekognition, upload taken photos, and compare those photos. We altered this code so that when the button is pressed, the Raspberry Pi automatically takes a picture, uploads it, and prints the result to the LCD screen.
Challenges I ran into
The biggest challenge we ran into was the Raspberry Pi micro SD card being corrupted about halfway into the project. We also had trouble getting the camera to consistently work and printing text to the LCD. The Camera always showed ENOSPC error where we had hard time enabling the camera component. Also, the LCD threw garbage values, largely because of loose connections, whenever no face was detected.
Accomplishments that I'm proud of
One of the biggest accomplishments was altering the GitHub software to print results to the LCD instead of to Raspberry Pi's desktop console. It finally printed the name on the LCD screen. It was accurate 10/10 times with even just one image per person in a training data set. It could also figure out the picture of the person stored in a phone when showed to camera.
What I learned
We learned a lot about programming in Python as well as using Raspberry Pi as a tool for powerful and simple project development. We also learned about Face Recognition system.
What's next for AlzLens
The logical next step for AlzLens is putting it in a sleek and compact package that is easy to use and test. The goal is to make a clippable device attached to the lens. Also, we have to deal with privacy issues.
Here is the one-minute promotional video: https://youtu.be/SROKXfRmHLU