Coastal areas are quite fragile landscapes as they are among the most vulnerable to climate change and natural hazards. In the case where the action of marine abrasion is greater than the deposit, there are evident cases of coastal erosion that have led to the disintegration and demolition of the earth’s surface. Coastline mapping and change detection are essential for safe navigation, resource management, environmental protection, and sustainable coastal development and planning.
Gigantum, decentralized, and user-friendly data science platform for POC development and collaboration. Jupyter Notebook web-based interactive computational environment for creating notebook documents where you can run and test code. Flask web framework for Python for developing and building AI/ML applications, Azure (clued infrastructure), and Github/Svelte front-end development for building user interfaces that will consume AI/ML models.
Coastline change detection from digital image data belongs to the boundary detection problem in the field of computer vision and image processing, in which edge detection and image segmentation are two conventional approaches to the boundary detection. In this particular use case, we utilized the Canny Edge Detection algorithm on satellite images acquired by the OLI (Operational Land Imager) sensor on Landsat 8 platform. Through this methodology, we were able to visualize and estimate the progress of the coastline erosion over time. The following open source packages have been employed: Rasterio to easily deal with raster images, OpenCV to apply the Canny algorithm, and Scikit-learn to segment images.
Implement and operationalize AI/ML algorithm to measure erosion, acquire image data from Landsat 8 platform, image data preparation (image standardization, RGB composite enhancement), image data segregation with K-mean clustering algorithm, model training with Edge Detection algorithm, candidate model operationalization, cloud deployment, and front-end development.