Aidan is a third year student at Washington State University, pursuing a B.S. in computer science. He grew up on Maui, spending time playing pickup basketball and looking for compelling new books. Aidan constantly strives to grow in his knowledge of machine learning and it’s wider fields. He is an intent collaborator, eager to lead and contribute in team activities such as Hackathon and Mentor Collective.

Home Island: Maui

High School: Kihei Charter High School

Institution when accepted: Washington State University

Akamai Project: Comparing the Performance of Semantic Segmentation Network Trained on Sequential Versus Single Frames

Project Site: Boeing Maui – Maui Space Surveillance Site – Kihei, Maui

Mentors: Trent Kyono, Jacob Lucas

Project Abstract:

The task of semantic segmentation is to identify and classify every pixel in an image. Using deep convolutional neural networks, we can achieve semantic segmentation over both individual and concurrent frames. We seek to determine whether the addition of sequential frames eases the training process. If so, it can be chosen as the preferred method between the two. Through the training of these multi-channeled networks, we modeled gradient descent optimization to quantify model accuracy. We then utilized an IOU (Intersection Over Union) evaluation metric to grade accuracy loss. In doing this, we determined the effectiveness of each of our models.