Spring 2017 Course Descriptions

COMP 150-02 Deep Learning for Computer Vision

G. Patterson
TR 7:30p-8:45p, Halligan Hall 111B
Q+ Block

This course investigates current research topics in data-driven object detection, scene recognition, and image-based graphics. We will examine large-scale datasets, convolutional neural networks, and several algorithms useful for understanding and manipulating visual data. These topics will be pursued through independent reading, discussions, student presentations, and projects involving current research problems. Students will train deep networks from scratch and fine-tune popular pre-trained networks. Students will gain experience with commonly used deep learning packages and learn practical techniques for network training. In the final project, students will implement and test their own improvements to state of the art networks.

Prerequisite: At least one of COMP 135 or COMP 136 or COMP 150-NLP.

Back to Main Courses Page