Designing A Deep Learning Based Backyard Squirrel Detection System Utilizing Raspberry Pi
Every year, my family looks forward to the month of May as our backyard donut peach tree is then filled with delicious peaches. I first noticed the backyard squirrel problem when I began remote learning due to COVID last year. Everyday, I saw my grandma working tirelessly in the yard watching for squirrels in an attempt to protect our peach tree. Over that summer, I attended an Artificial intelligence research program and discovered the potential of Artificial intelligence. As a result, I decided to apply what I learned to a real world problem that was relevant to those like my grandma. I designed and constructed a system which autonomously detected passing squirrels and sent out alerts. My main device consisted of a Raspberry Pi 4B, a Google Coral Edge TPU accelerator, and a miniature camera. In order to detect the squirrels, I retrained a deep learning model by using individually collected squirrel image datasets which varied in size, position, and color. Upon testing, my device produced significant results, detecting squirrels with outstanding accuracy and confidence, with no false detections.