Satellite Image Processing and Analysis in the Big Data Era
Course content
We live in exciting times with a rapidly expanding number of earth observation products available to understand global environmental and socio-economic changes. However, this development poses several challenges:
- How to process large remote sensing data sets?
- How to extract and analyze patterns and trajectories of land change?
Through this course, the principles of big data analysis will be introduced. This course emphasizes advanced machine- learning and pattern recognition, for instance, convolutional neural networks, random forests and other classification techniques. We will take a deep dive into how to evaluate machine learning models' performance and assess the accuracy of produced classification results. We will also work with Google Colab, PyTorch, Python-based libraries for GIS, and cloud-image processing platforms like Google Earth Engine to process and analyze various remote sensing data sets. The students are expected to progress their scripting skills in a Python/ Javascript environment.
The course welcomes those who would like to advance their knowledge in remote sensing of the environment and socio-economic footprint, pattern recognition and spatial data analysis and who would like to advance their skills in machine learning. However, this class is also open to those who already have skills in programming and machine learning, but seek to extend their portfolio in satellite remote sensing applications.
MSc Programme in Geography and Geoinformatics
Knowledge:
- Sources of satellite data; currently operating non-commercial and commercial platforms in Earth Observation, with a specific focus on very-high-resolution imagery,
- Principles of pattern recognition, classification and segmentation using convolutional neural networks (CNNs),
- Advanced satellite image processing and classification in a cloud environment,
- Python Application Programming Interface (API); API-based computation with Python for big data processing.
Skills:
- Hands-on experience in parameterization, running and evaluation of the performance of convolutional neural networks for pattern recognition tasks using, e.g., Google Colab and PyTorch,
- Progression of skills in Python-based analysis of satellite data using machine learning,
- Ability to download, pre-process, and analyze large amounts of satellite imagery with a focus on remote sensing of the Geobiosphere and Land System Science.
- Advanced use of cloud-based classification via Google Earth Engine (GEE), and Google Colab.
Competencies:
-
Ability to process and analyze various types and a large amount of remote sensing data sets using advanced state-of-the-art machine learning and cloud-based methods.
The form of teaching is theory exercises combined with lectures and various forms of activation of learning. For the teaching plan, please see Absalon.
Please see Absalon course page.
BSc in Geography and Geoinformatics or equivalent. Prior
experience in satellite remote sensing processing and analysis is
expected. Experience in scripting is welcomed.
The course builds on prior knowledge of working with satellite
imagery, such as passing via Bachelor level classes such as
Introduction to Remote Sensing, Classification of Spatial Data, as
well as some Master-level remote sensing classes, such as Remote
Sensing of Geobiosphere, Remote Sensing in Land Science Studies,
Spatial pattern analysis. However, we will try to accommodate the
students without prior experience in remote sensing but with solid
experience in programming machine learning. If you dit not take
remote sensing classes, please contact the course responsible prior
to enrolling in the class.
- ECTS
- 7,5 ECTS
- Type of assessment
-
Written assignment, During courseOral examination, 20 minutes
- Type of assessment details
- A set of lab reports developed through the course must be handed in prior to the exam week. The oral exam uses the lab reports as its point of departure. It includes the titles listed in the officially approved reading list. A combined grade is given after the oral exam.
- Aid
- All aids allowed
- Marking scale
- 7-point grading scale
- Censorship form
- No external censorship
Several internal examiners
- Re-exam
-
Identical to the ordinary exam.
If the quality of the Lab reports are not acceptable, the student can choose to either hand in a new or revised report.
If the quality of the Lab reports are acceptable, the student can either hand in a revised report or resubmit the original report from the ordinary exam.
The Lab reports must be handed in prior to the re-examination week. The oral exam uses the written assignment as its point of departure. It includes the titles listed in the officially approved reading list. A combined grade is given after the oral exam.
Criteria for exam assessment
See learning outcome.
Single subject courses (day)
- Category
- Hours
- Preparation
- 171
- Theory exercises
- 35
- English
- 206
Kursusinformation
- Language
- English
- Course number
- NIGK22000U
- ECTS
- 7,5 ECTS
- Programme level
- Full Degree Master
- Duration
-
1 block
- Placement
- Block 4
- Schedulegroup
-
A
- Capacity
- 22
The number of places might be reduced if you register in the late-registration period (BSc and MSc) or as a credit or single subject student. - Studyboard
- Study Board of Geosciences and Management
Contracting department
- Department of Geoscience and Natural Resource Management
Contracting faculty
- Faculty of Science
Course Coordinator
- Alexander Prishchepov (4-65707476446d6b72326f7932686f)
Teacher
Ankit Kariryaa, Alexander Prishchepov, Xiaoye Tong, Rasmus Fensholt
Er du BA- eller KA-studerende?
Kursusinformation for indskrevne studerende