• KJ Ha

ML4PCOMP Assignment 4

I have two separate ideas for the final project.

Similar to the project I did for Intro to Pcomp class, I would like to build a simple watering device using a teachable machine algorithm. The device will allow users to give a simple voice command to activate the water pump that is connected to Arduino. I can imagine that this project won’t be coding heavy; more weight would be put into the serial communication part.

However, I’m currently more interested in looking into machine learning architecture than giving it a physical output so, there comes my second idea. The second option for the final project is to keep working on improving the existing model. I would like to dig deeper into FruitToEmoji and to run a whole process again in my own way, from collecting data to re-design the layers then hyperparameter tuning.

If I end up pursuing this second option, I would like to create datasets for five different colors (white, blue, green, yellow, and red) and if each color is detected, the corresponding color of LED will light up. I'm also considering augmenting the data to improve the performance.