Deep Learning for Landcover Classification

The number of data produced by sensing devices has increased exponentially in the last few decades, creating the “Big Data” phenomenon, and leading to the creation of the new field of “data science”, including the popularization of “deep learning” algorithms to deal with such data. In the field of remote sensing, the number of platforms for producing remotely sensed data has similarly increased, with an ever-growing number of satellites in orbit and planned for launch, and new platforms for proximate sensing such as unmanned aerial vehicles (UAVs). Fortunately, the increase in the number and heterogeneity of data sources (presenting both challenge and opportunity) has been paralleled by increases in computing power, by efforts to make data more open, and by advances in methods for landcover classification and data fusion. Deep learning has been used intensively in the remote sensing community for landcover classification using both single and multisensory data.

In this seminar, groups of students will each present an approach in lecture and elaboration by emphasizing the use of deep learning for landcover classification from the perspectives of single and multisensory techniques. Students also apply a number of deep learning-based classifiers on real satellite images. For this purpose, we will distribute a proper amount of codes among the students for benchmarking and evaluation.

Here, the important dates are listed:

§ Email to me for your task (Nov. 12, 2020)

§ I share a link to a remote sensing data set (Nov. 12, 2020)

§ Set Skype meetings with me (Nov. 30, 2020)

§ Mid-term presentation (Dec. 4, 2020) (5 minutes per person followed by 5 minutes Q&A)

§ Final presentation (Feb. 5, 2020)

§ Submission of the seminar paper (Feb. 25, 2020)  (6-8 pages per person) (I share a template)