A.I. Powered Pet Feeder

Presenter(s)

Ian Pfenning

Abstract or Description

This project’s purpose is to design and create a pet feeder that uses artificial intelligence (A.I.) to detect two different animals, and respond based on the needs of each individual pet. This eases the workload of a human with multiple pets, especially when said pets have different eating needs or habits. This project uses A.I. hosted independently in Python. The program takes live input from an external webcam, runs the image through the A.I. model, and outputs through an external LCD screen. The project uses Teachable Machine to host the A.I. model during training. The model uses approximately four hundred and fifty images in the model, with about 200 images of each dog. Currently, the project is limited to two different dogs, and does not yet include a physical feeding system. However, the project reliably identifies images as one of three classes: two different dogs and not a dog. Additionally, the program stores data of how many times each dog has been detected, simulating each time a dog has received food from the dispenser, and allows a limited number of times each dog can be fed. The project is currently in a working prototype state.