Effectiveness of Super Resolution Convolutional Neural Network in Enhancing Land Cover Classification from Mid-Resolution Satellite Images

10/04/2017
1:30 pm - 4:00 pm
Expo Hall/Career Fair

Objective: Inspiration
Audience Level: All
Session Type: Poster

Remote sensing data provides essential information for monitoring the earth. But, till date, a major constraint is low spatial resolution. We investigated the effectiveness of Super Resolution Convolutional Neural Network to upscale the satellite image. We succeeded to classify features that are few pixels across and are misclassified under the same classification method but without up scaling.

Speaker(s)

Debolina Halder, Student, Bangladesh University of Engineering & Technology