Skip to content
Nathan Rosenberg edited this page Feb 20, 2021 · 4 revisions

KOLT (known object localization and tracking) combines a deep convolutional neural network with a tracker and Kalman filter and is able to localize and track multiple objects at once. It is designed to be a solution for new roboticists at completing their MQPs at WPI that require a vision solution as part of their project. Because creating a computer vision system is very time consuming and is usually not in itself the final product for an MQP, the goal of KOLT is for it to empower these projects and allow them to be more successful by giving them more time to work on the rest of their project.