• home
• current students
• graduate
• events

Presenter: Ahmed Elmougi
Supervisor:

Date: Thu, March 4, 2021
Time: 12:00:00 - 13:00:00
Place: ZOOM - Please see below.

ABSTRACT

Zoom meeting link: https://uvic.zoom.us/j/84665707506?pwd=MkZkbVlUVGtyOHc5MSt1L1I5YTkvQT09

Meeting ID: 846 6570 7506

Password: 463112

Note: Please log in to Zoom via SSO and your UVic Netlink ID

Abstract: Neural networks are a powerful tool for many data applications. However, its high dimension nature, large network size, and computational complexity, and the need for large amounts of training data make it challenging to be used in edge computing applications, which are becoming increasingly popular, relevant, and important. We propose using SURF descriptors instead of the original image as input to both convolutional neural networks (CNN)  and recurrent neural networks (RNN) to learn the pose (position and orientation) of the input image in indoor environments. The results show that we can reach comparable performance to the state-of-the-art complex neural network methods that use full images as input with orders of magnitude reduction in the training time and storage space.