Introduction Praveen #98
praveen-86176
started this conversation in
General
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
Praveen Kumar
Introduction
Hi! I am Praveen, a 2nd year Computer Science engineering student from Bangalore, India,
with hands-on experience in Python and Machine Learning/Deep Learning. I am interested in
contributing to the Automated Coastline Extraction for Erosion Modeling in Alaska
project as part of GSoC 2026.
Research I've Done
After exploring the CoastlineExtraction repository,
I have a clearer understanding of the current pipeline — particularly the use of NDWI-based
thresholding with a majority sliding window for water detection, and the challenges faced
when applying the DeepWaterMap model (originally trained on 30m Landsat imagery) to
higher-resolution PlanetLabs imagery (3–5m). I find the retraining approach using
auto-labeled PlanetLabs data particularly interesting, and I believe there's significant
scope for improvement in cliff area segmentation and handling challenging conditions like
water shadows and satellite artifacts.
I also went through the potential improvement areas listed in the project idea and I'm
especially drawn to SWIR and Elevation Data Integration and Improved Cliff Area
Segmentation, as these seem to be the most impactful from a modeling perspective.
Why This Project?
Arctic coastal erosion is one of the most urgent and underrepresented climate challenges
of our time — directly threatening Alaskan communities. The intersection of deep learning,
remote sensing, and real-world climate impact makes this project genuinely meaningful to
me, and I'm excited to contribute toward building a more accurate and automated solution.
My Social Links
Beta Was this translation helpful? Give feedback.
All reactions