GeoAI for data accuracy – seminar

Methods of Geospatial Artificial Intelligence (GeoAI) have developed rapidly. Concurrently, there is a rising demand for greater accuracy in both data and products. Artificial intelligence methods can help us to meet these needs.

In the context of the National Land Survey of Finland (NLS), the AI4TDB project aimed to enhance the accuracy of the topographic database using AI methods in the year 2023. This project primarily focused on improving the representation of buildings and watercourses. Throughout the project, we conducted numerous experiments and gained substantial knowledge in several areas:

a. Techniques for enhancing the accuracy of building vectors in the topographic database.
b. The utilization of deep learning for processing aerial images and Lidar data.
c. Enhancing watercourse detection through the incorporation of extended information.
d. Leveraging deep learning methods for monitoring changes in land use.
e. Deployment strategies for practical use of AI models in map production, among other topics.

In this event, we are honored to invite Dr. Petri Rönnholm, Senior University Lecturer, in Department of Built Environment, Aalto University, as the keynote speaker. He will talk about Geospatial Artificial Intelligence. If you’re interested in the topics and would like to dive deeper, we cordially invite you to register for the event.
This event is organized by NLS and Geoforum. The event will be organized as a hybrid, allowing you to participate either in person at Maria01 in Helsinki or online. The seminar is scheduled for December 20, 2023, from 13:00 to 15:45. 

Registration

Register to participate: https://www.lyyti.in/seminar_GeoAI

The registration for the in-person event closes on December 13, 2023. The registration for the online event closes on December 18, 2023. Teams participation link will be sent to registered participants two days before the event

Programme

• 13-13.10 Heli Laaksonen: Introduction
• 13.10-13.40 Petri Rönnholm: Keynote speech: Geospatial Artificial Intelligence
• 13.40-13.55 Jere Raninen:  AI application: Improving the positional accuracy of building vectors in topographic database 
• 13.55-14.10 Pyry Kettunen: Creating topographic watercourse network using computer vision
• 14.10-14.25 Christian Koski: Extraction of hydrographic features from remote sensing data using deep learning
• 14.25-14.35 Break
• 14.35-14.50 Emilia Hattula: Comparison of building detection in the forest environment using UNet for Images and Lidar data
• 14.50-15.05 Emilia Söderström: Monitoring changes in land use using deep learning methods
• 15.05-15.20 Eemeli Pettersson: How to use the AI models in the open-sourced platform?
• 15.20-15.35 Lingli Zhu: A summary of AI projects in the National Land Survey of Finland 
• 15.35-15.45 Heli Laaksonen: Questions and discussions