2024-2025 Academic Catalog

Jul 03, 2024

2024-2025 Academic Catalog

DSCI 215 - Applying Deep Learning to Earth Observation

Focusing on convolutional and recurrent neural networks, this course will provide an in-depth overview of key machine learning algorithms and their application to satellite imagery (especially for the task of semantic segmentation), including the full workflow required to acquire and process imagery, develop and train a model, and make and critically evaluate the resulting maps. The course will be strongly hands-on, and emphasize the use of programming (python), open image archives and EO analytical platforms (e.g. Google Earth Engine), and tools for creating open and reproducible workflows (git and GitHub).

For DSCI students, this course satisfies the capstone requirement.

Prerequisites: There are two sets of prerequisites for this course, one to ensure that students have sufficient background in GIScience and the other to ensure that they have enough programming experience.

GIScience Prereqs:
GEOG 293 or GEOG 383 with preference given to students that have taken GEOG 282 or GEOG 382 and GEOG 296 or GEOG 397

Programming Prereqs:
Can be satisfied through two approaches -

Approach 1 - take at least one: DSCI 105 or DSCI 305 ; DSCI 125 or DSCI 304 ; CSCI 120 or CSCI 301

Approach 2 - take at least two: SSJ 302 ; SSJ 30274 ; GEOG 246 or GEOG 346

Determination regarding whether prerequisites can be waived or satisfied by other means may also be made by the instructors.

Anticipated Terms Offered: Annually