Welcome to CoastlineExtraction Discussions! #12
Replies: 31 comments 36 replies
-
|
Hello, |
Beta Was this translation helpful? Give feedback.
-
|
Warm Greetings, |
Beta Was this translation helpful? Give feedback.
-
|
Hello @fwitmer, while reviewing the Coastline Extraction repository, I've been familiarizing myself with the function of each file. I have a query regarding obtaining a sample dataset to initiate my work. Could you kindly suggest some data sources for this purpose? Thank you very much! |
Beta Was this translation helpful? Give feedback.
-
|
Thanks very much to everyone for your interest in this project! To help get you @Aashish-100 get started, I added some sample imagery to the repository. If you would like to explore and download additional PlanetLabs imagery, you can request access here: https://www.planet.com/markets/education-and-research/. |
Beta Was this translation helpful? Give feedback.
-
|
Greetings, |
Beta Was this translation helpful? Give feedback.
-
|
Hello @fwitmer. Hope you find my message. Would love to hear from you. |
Beta Was this translation helpful? Give feedback.
-
|
Thank you very much for the assistance @fwitmer. Will get back to you as I keep making updates on the project. |
Beta Was this translation helpful? Give feedback.
-
|
Greetings, I'm Shashank Shekhar Singh, a second-year student at IIT BHU, India. My passion lies in harnessing the power of Data Science, Machine Learning, and Deep Learning to address pressing environmental issues like implementing geo-spatial data science for satellite image analysis. I am eager to contribute to the project on Automated coastline extraction for erosion modeling in Alaska. I would be highly grateful for your guidance on where to start. Currently, I am immersing myself in understanding the algorithms, functions and its applications. Thank you sincerely, Shashank Shekhar Singh |
Beta Was this translation helpful? Give feedback.
-
|
Hello everyone , I'm Ayush Kumar Gaur, a first-year Btech computer science enthusiast aims to research on the captivating world of AI and ML to harness its full potential and to apply it in various fields i am proficient in python and understanding maths behind machine learning models such as linear regression and decision trees and also have a grasp over python libraries such as numpy, matplotlib and pandas in order to deepen my understanding in this field i wish to work and contribute in real world projects with an interactive community and understand dynamics of real world problems like Global warming to improve myself currently i am learning python libraries such as rasterio and rastertools and other libraries that were unknown to me and used in this project ,to understand to code and contribute to it as well. Thank You |
Beta Was this translation helpful? Give feedback.
-
|
Hey @fwitmer, I have been fidgeting around with the codebase for a while. I came across the ndwi_labels.py file, figured out what's going on. I was thinking how about trying out the EWI (Enhanced Water Index) or MNDWI(Modified Normalized Difference Water Index) algorithms instead of NDWI. Considering the terrain, EWI might outperform MNDWI at some places. Later on, we might design and train a model architecture capable of accurately segmenting coastlines from satellite imagery. Kindly let me know what you think about it! |
Beta Was this translation helpful? Give feedback.
-
|
@fwitmer please let me know what you think about it! |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for those suggestions. It is certainly worth considering other indices that might help, but the PlanetScope imagery has limited spectral bands, so I don't think the EWI or MNDWI are possible to calculate. |
Beta Was this translation helpful? Give feedback.
-
|
@fwitmer greeting one and all i am Suvrat Patel I have my Bachelors Degree In Artificial Intelligence And Machine Learning Currently I am Writing my Research paper as well as i have experience in implementing research paper project to real go, i have worked on similar projects with similar skill set i have skills on python, Machine Learning, Deep learning, Geospatial data Processing as well as Satellite imagery, in past i have worked on similar project with ISRO(Indian Space Research Organization ) Satellite imagery which is real time based project for Human Lives. |
Beta Was this translation helpful? Give feedback.
-
|
Greetings to @fwitmer and all, my name is Ruoyu Li and I am an incoming grad student of Information Systems at Northeastern University in USA this fall. I am very passionate to work in the coastline extraction project. I am proficient in python and machine learning. I also have experiences in geospatial data processing. Meanwhile, as a former landscape architect, I care a lot about the ecosystem and I completed many related projects both in design and coding(such coral reef resistency design against the climate change, water and terrain analysis with python etc). Among those projects, they all have research involvement as I really enjoy the combination of research and practices. I really want to contribute to this project as it matches my professional pursuit - to raise more social awareness about ecosystem via technology. I am currently going over the documentation and get familiar with the files.\ Thank you |
Beta Was this translation helpful? Give feedback.
-
|
Greetings |
Beta Was this translation helpful? Give feedback.
-
|
Yes, using a U-Net CNN and/or other similar methods makes sense. |
Beta Was this translation helpful? Give feedback.
-
|
Hello @fwitmer Sir! I am a junior in a Computer Engineering course here at JNU, New Delhi. I feel quite strongly for active climate control and my efforts to reverse the or contain the damage in any form. I've drawn inspiration to work in this domain from a recent talk by a Yale Climate Fellow in our college. Point is my motivations to work on this project remain crystal clear, as for the project needs - As I understand you have already done the following:
Also, I’d like to highlight my previous work in the ML/AI, where I’ve done considerable work in a number of different domains, including: Reinforcement learning, NLP-LLMs(current) and Supervised ML Projects. After starting with the code base review, I came across your preprocessing methods and would like to add, if we implemented Elastic Deforming, Color jittering and Random Rotating, we'd be closer to achieving our end goal of a more clear and precise model. I've so far been trying to implement this on EuroSat dataset to confirm my theory. I've also sent a request for the original planetLabs dataset and as soon as I get it, I'll try to generate a PR so you can check the work done. |
Beta Was this translation helpful? Give feedback.
-
|
Hi 👋 @fwitmer and everyone interested in the Coastline Extraction project! I'm Wayne Zhou, a Master's student in Computer Science at Northeastern University. Previously, I worked as an architectural designer specializing in building information modeling (BIM), GIS, and green technology. I'm passionate about projects with real-world impact and problem-solving potential. Outside of coding, I enjoy classical arts, modern furniture, and exploring nature. I've been reviewing the repository and have a few questions to get started:
Thanks in advance for your time and guidance! |
Beta Was this translation helpful? Give feedback.
-
|
Hello everyone! Looking forward to learning more about the project and exploring how I can contribute! |
Beta Was this translation helpful? Give feedback.
-
|
Hi, |
Beta Was this translation helpful? Give feedback.
-
|
Hi @fwitmer, @pradeeban, and everyone, I’m Varun, a CS undergrad, and I came across this project through GSoC 2025 and found the idea of using satellite imagery for coastal erosion analysis really interesting! I’ve been exploring NDWI visualization and would love to contribute, possibly by improving error handling for missing/invalid tiles. Is this forum still active? Also, does the project have a Slack or another communication platform for discussions? Looking forward to contributing! Thanks! |
Beta Was this translation helpful? Give feedback.
-
|
@pradeeban Thanks for the response! I appreciate the clarity on keeping discussions open here. I was looking into potential ways to contribute and was thinking to start by improving error handling for missing or invalid tiles. Would this be a meaningful contribution? Do missing tiles currently cause issues in the coastline extraction pipeline, or is it something that’s already well-handled? If there are any specific areas where contributions are most needed, I’d love to get started on them! Looking forward to your thoughts. |
Beta Was this translation helpful? Give feedback.
-
|
@varun-aahil, Mentors (@fwitmer and @rawann31) for the project will have a better answer for your question since I am not familiar with this project enough. But as org admin, I could give you some general suggestions (which are actually true for all the organizations, not just Alaska): |
Beta Was this translation helpful? Give feedback.
-
|
Thanks, @pradeeban! I really appreciate the advice and will definitely keep that in mind while contributing. @fwitmer, @rawann31, since there aren’t any open issues, I’ll start exploring the codebase and documentation to better understand the extraction pipeline. If I find areas that could use improvement, I’ll open an issue or a PR. I’ll reach out if I have any questions along the way. |
Beta Was this translation helpful? Give feedback.
-
|
Hi @rawann31, I hope you're doing well. Just wanted to follow up on the pull request I submitted addressing the inconsistent date format issue in Additionally, I plan to raise more PRs as I continue exploring and improving the codebase. It would be great if you could check them from time to time. Thank you! |
Beta Was this translation helpful? Give feedback.
-
|
@rawann31 Can I email you my proposal, if yes can you please check it out |
Beta Was this translation helpful? Give feedback.
-
|
Beta Was this translation helpful? Give feedback.
-
|
Hello Everyone! |
Beta Was this translation helpful? Give feedback.
-
|
Hello @rushichavda Thanks for your interest in the project! If you’d like to contribute, you’ll need to submit a strong GSoC application that highlights your experience with geospatial data, deep learning, and frameworks like PyTorch or TensorFlow. |
Beta Was this translation helpful? Give feedback.
-
|
I have worked on raster data, from google earth engine, with several processing libraries. i have also used geospatial foundational model prithvi-100 - i have finetuned it. i have developed pipeline to process the satellite images. |
Beta Was this translation helpful? Give feedback.
-
|
would this experience help? this i did in my internship at IBM research. |
Beta Was this translation helpful? Give feedback.
-
|
Hey everyone, Sahil Chordia here! 👋 Really excited about the possibility of participating in GSoC 2025, and I'm super interested in projects that leverage AI, machine learning, and automation. I am a student at the University of Maryland, pursuing a Master's in Data Science, I'm eager to explore how we can apply these technologies to enhance coastline extraction methods, maybe through more efficient data processing or automated analysis. I'm really looking forward to hearing your thoughts, learning from all of you, and brainstorming some ideas! Cheers, |
Beta Was this translation helpful? Give feedback.
-
|
Hey everyone, I'm Prasanna. I’ve been going through the CoastlineExtraction repo lately and got really interested in the Planet imagery + NDWI workflow. I’m hoping to understand the pipeline better and start helping out, especially around improving the segmentation side of things. Looking forward to learning from you all and contributing where I can. |
Beta Was this translation helpful? Give feedback.
-
|
Hi mentors (@fwitmer), My name is Lakshya, and I am a student developer interested in contributing to the Automated Coastline Extraction project. I have a background in Python and I am currently ramping up on the geospatial stack (rasterio, geopandas). I have successfully set up the local environment and have written a simulation script to replicate the NDWI (Normalized Difference Water Index) logic mentioned in the project overview to understand how the water masking works. You can see my NDWI reproduction script here: https://github.com/lakshyadchad/alaska-gsoc-prep I am now moving on to testing the pipeline with the sample PlanetLabs imagery provided in the repo. I am particularly interested in the "Cliff Area Segmentation" challenge and will be looking for guidance on where I can be most helpful with testing or documentation in these early stages. Looking forward to contributing! |
Beta Was this translation helpful? Give feedback.
-
|
@lakshyadchad Thanks for your interest, Lakshya! We just updated the master branch, so please be sure to take a look at the latest version of the code base. |
Beta Was this translation helpful? Give feedback.
-
|
Okay @fwitmer sir, I will look at them. Thank you for updating me. |
Beta Was this translation helpful? Give feedback.
-
|
Hi @fwitmer sir and mentors, Do you have any suggestions on how to proceed or alternative options to look for? |
Beta Was this translation helpful? Give feedback.
-
|
Hello @fwitmer , I'm Prasanna, interested in GSoC 2026. I talked with @Ritika-K7 about a month ago and she suggested I go through the codebase and get familiar with it. I've been exploring the dev branch and tried running the pipeline locally. I noticed the project doesn't have automated tests yet. I think adding testing infrastructure would be useful - it makes it easier to verify changes work and helps future contributors feel confident their code won't break things. So I've started building a testing infrastructure for the project. I've set up pytest framework with basic fixtures and added 9 tests for config loading (all passing).
Branch: https://github.com/PrasannaPal21/CoastlineExtraction-gsoc/tree/testing-infra Would appreciate any feedback - is this useful? Any specific parts you'd want tested first? |
Beta Was this translation helpful? Give feedback.
-
|
UPDATE - Progress: pytest setup , config tests , CRS tests https://github.com/PrasannaPal21/CoastlineExtraction-gsoc/tree/testing-infra |
Beta Was this translation helpful? Give feedback.
-
|
Quick Update: Just added 16 tests for NDWI calculations and raster tools. They check everything from sliding window setup and noise cleanup (via morphological transforms), to k-means clustering, the actual NDWI math, and Otsu thresholding for clean water/land splits. Why i added these: NDWI is the heart of our water/coastline detection from satellite images - positive values flag water, negatives mean land. Passing these tests means no math bugs that could mess up coastlines later. Progress so far: pytest basics, config checks, CRS handling, and NDWI. Branch: github.com/PrasannaPal21/CoastlineExtraction-gsoc/tree/testing-infra |
Beta Was this translation helpful? Give feedback.
-
|
hello this ensures satellite images actually overlap with shapefiles before heavy processing so no wasted compute on misaligned data. Progress: pytest, config, CRS, NDWI, spatial utils. branch - https://github.com/PrasannaPal21/CoastlineExtraction-gsoc/tree/testing-infra |
Beta Was this translation helpful? Give feedback.
-
|
Hi @fwitmer sir, I added tests for NDWI/raster stuff, spatial checks like overlaps and geometry, and some integration tests that run the full pipeline on real Planet satellite images. I tried these on a sample input, all the tests were passing. the tests I added were from my knowledge of testing, would like to know your view on them if they make sense in the project or not. This means we catch math bugs, data mismatches, and workflow glitches early, super important for reliable coastline extraction without wasting time on bad data. things covered: setup, config, CRS, NDWI, spatial utils, integrations, docs, and requirements. checkout the testing module here - https://github.com/PrasannaPal21/CoastlineExtraction-gsoc/tree/testing-infra @fwitmer, @rawann31, @Ritika-K7 I would appreciate your views on the testing infrastructure for the project. |
Beta Was this translation helpful? Give feedback.
-
|
Hi @fwitmer, |
Beta Was this translation helpful? Give feedback.
-
|
Jiya, thanks so much for bringing this to our attention. I just requested that LJitendra05 modify their PR which tried to fix this issue. The dev branch does exist, so hopefully you are able to access it here: |
Beta Was this translation helpful? Give feedback.
-
|
Hi @fwitmer , I'm Samta Sakha, a Data Science and AI student from India, and I'm very interested in the Automated Coastline Extraction for Erosion Modeling. I have been exploring the repository and understanding the workflow around NDWI-based labeling and the DeepWaterMap training pipeline. While reviewing the preprocessing steps, I noticed that Planet imagery provides a Usable Data Mask (UDM) that identifies clouds, shadows, and other low-quality pixels. Integrating the UDM mask during preprocessing could help filter unusable regions, improve the quality of NDWI-generated labels, and reduce noise in the training dataset. I opened an issue #89 regarding this and would be interested in exploring how UDM masking could be incorporated into the preprocessing pipeline. Please let me know if this would be a useful direction to explore, and I would be happy to contribute to its implementation. |
Beta Was this translation helpful? Give feedback.
-
|
Hello, I am Menna Ezz Elarab, a senior year Computer Engineering student from Ain Shams University in Egypt. I have skills in Python, machine learning, and deep learning. I haven't contributed to any open source projects before, but I am willing to take on a new challenge this year. This summer, I am interested in contributing to Project Idea 1: Automated coastline extraction for erosion modelling in Alaska, as it aligns with my previous experience with a water segmentation project. It was my first experience with working on multi-band satellite imagery and geospatial data processing. You can find this project here: (https://github.com/MennaEzzelarab/water_segmentation/tree/main). |
Beta Was this translation helpful? Give feedback.
-
|
Hello @fwitmer, |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for everyone's interest in our project! We are excited to start reviewing applications. If you are still wondering where to focus your efforts, you might think about how you can improve the automatic generation of training/labeled data. That is what we are trying to do with the NDWI sliding window, but it needs improvement. Once we have a robust set of training data, then the various models should do much better. Suggesting improvements/ideas on the modeling side are also welcome. Good luck everyone! |
Beta Was this translation helpful? Give feedback.
-
|
Hi everyone, I'm Romit Basak, a first-year MS Computer Science student at Northeastern University (Khoury College), working in GeoAI and remote sensing ML. I'm interested in applying to Project 1 (Automated Coastline Extraction) for GSoC 2026. My relevant experience comes from developing a satellite-based water detection and flood prediction system for urban environments. I've worked extensively with multi-band satellite imagery including Sentinel-1 SAR and optical sensors, NDWI/MNDWI thresholding for water body detection, and pseudo-label generation using optical imagery to supervise SAR-based models — a workflow closely related to the NDWI-with-majority-sliding-window approach described in this project. I've also worked through the specific failure modes of threshold-based water detection in challenging environments: false positives from radar shadow in dense urban SAR, building-shadow contamination in optical NDWI, and the class imbalance problem in rare-event detection. I looked through the CoastlineExtraction repository and have a few questions:
Happy to take on any starter tasks. I'm familiar with the geospatial Python stack (rasterio, GDAL, GEE, PyTorch) and have experience with U-Net-style segmentation on satellite imagery patches. Looking forward to the conversation. Romit Basak |
Beta Was this translation helpful? Give feedback.
-
|
Hello @fwitmer and @Ritika-K7, I am currently finishing my proposal for GSoC 2026. I have a few questions.
Thanks in advance. |
Beta Was this translation helpful? Give feedback.
-
March 31 - 18:00 UTC April 21 - 18:00 UTC |
Beta Was this translation helpful? Give feedback.
-
|
okay no problem. Thank you |
Beta Was this translation helpful? Give feedback.
-
|
@pradeeban If I choose to extend the time and add 2 additional months, should I ignore the midterm and final evaluation dates mentioned in the GSoC program timeline, or should I add them to my timeline as they are? |
Beta Was this translation helpful? Give feedback.
-
|
Then you should add the updated timeline in your proposal. The mid-evaluation will come right in the middle as your extended timeline. So, if you extend it by 8 weeks, your mid-evaluation comes 4 weeks later than the original mid-evaluation time. |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for your help. |
Beta Was this translation helpful? Give feedback.
and others
Uh oh!
There was an error while loading. Please reload this page.
-
👋 Welcome!
We’re using Discussions as a place to connect with other members of our community. We hope that you:
build together 💪.
To get started, comment below with an introduction of yourself and tell us about what you do with this community.
Beta Was this translation helpful? Give feedback.
All reactions