Applied and Computational Math Seminar: Convolution Networks for Object detection in Satellite Imagery

Speaker:Mahmoud Lababidi, DigitalGlobe
Title: Convolution Networks for Object detection in Satellite Imagery

Abstract: While the production of research in image based neural network research has been accelerating at a rapid speed, many of these techniques are being refitted into the space of satellite imagery, where some advantages and disadvantages come to light in working with these neural networks. We will delve into understanding satellite images and their various shapes and sizes and the progress of machine learning techniques used to perform object detection, segmentation and some aspects of unsupervised learning approaches for object similarity and the image enhancement technique, super-resolution.

 

Time: Friday, September 29, 2017, 1:30-2:30pm

Place: Exploratory Hall, Room 4106