What this project is

Even though satellites are being utilized more frequently to monitor coastal areas, conventional techniques mainly depend on comparing images manually and using large aggregates over comparatively low frequencies of periodical assessments. Such an inconsistency does not allow to detect erosion at early stages and predict it soundly in time.

The present research project is based on my master’s thesis devoted to the observation of coastline erosion in three designated beaches of Corfu Island, Greece. The goal of this project is to create CoastGuard AI, an open-source system that combines old and current space images and uses modern AI technology to detect erosion processes. The main hypothesis of the research pertains to whether the application of computer programs performing comparison of satellite images from many years ago with contemporary materials can allow detecting possible erosion trends adequately enough to take efficient preventive measures.

For the project initiation, we have planned to do the following activities:

1. To create a detection framework. We will combine computer vision and modern techniques of deep learning with free satellite imagery (Sentinel, Landsat, historical aerial photographs taken in the last 20-30 years) for detecting the processes of shoreline changes.

2. To test the obtained results against ground truth. Verification of the obtained results happens during the comparison of the identified shoreline changes with the field measurements and historical information for the three sites of Corfu, distinguishing actual marks of erosion from other imprints obtained in the course of imaging (influences of tides, seasonality, clouds).

3. To create a risk map. We will create a risk map of erosion points in the three coastal zones.

4. To publish the results obtained for re-using by others. All methods and developing pipeline will be published for other Greek coastal regions and for the application of the same methods and means in other countries

The project will not include any direct measures concerning the coasts (construction or interventions)- it is limited only to detection, measuring, and forecasting.

Theory of impact

Coastal erosion is most often recognized after visible damage has occurred, e.g., when a beach has become visibly narrow, a road or construction has been endangered, protective systems have collapsed. By the time this becomes apparent without the need for technological supervision, the main chance for intervention has already been missed.

Greece's municipalities and engineering agencies still lack readily attainable, cost-effective, and repeatable tools to monitor changes in erosion as it progresses along coastlines. My thesis illustrated that even a few well-researched coastlines require extensive manual analysis which hinders the number of coastal areas that are monitored successfully.

CoastGuard AI resolves the problem through making the shoreline changes detection automated and repeatable based on the free satellite data instead of carrying out expensive individual surveys for each site. As a result, it offers the local authorities a tested pilot for three real and well-studied coastlines in Corfu which enables them to obtain significant and quantifiable data on erosion that has not been accessible so far.

The longer-term objective is to normalize this sort of monitoring first throughout other Greek coastlines and subsequently throughout Europe, where similar erosion pressures occur, but the capacity for monitoring is rather limited. If the project proves successful, it would serve as a guide for other regions to follow without having to invent a methodology for detecting erosion.

How the money will be spent

Minimum Budget: €12,000

€3,000 — Computational power (GPU/cloud) trained AI detection model on satellite imagery.

€2,500 — Field validation: travel, apparatus, and ground truth measurements from 3 Corfu coastline locations.

 €3,500 — Basic platform/dashboard design with detected erosion and risk map presentation.

 €1,500 — Contingency (extra data requirements, delays).

 €1,500 — Personal time allotment for project implementation with ongoing employment.

Ideal Budget: €30,000 (based on minimum)

 €2,000 — Access to premium/historical satellite archives where free data (Sentinel, Landsat) isn't sufficient.

€6,000 — Part-time AI/ML collaborator support (approx. 4-5 months), to strengthen the technical depth of the model beyond what I can build alone.

€3,000 — Part-time GIS/data engineering support, so validation and pipeline work can proceed in parallel.

€2,500 — Public-facing platform with additional features, rather than a minimal internal dashboard.

€2,000 — Extended validation to a fourth or fifth coastline beyond the initial three.

€2,000 — Additional buffer for scaling and unexpected costs.

Who is on your team? What’s your track record on similar projects?

My name is Giorgos Tsimpoulis and I am a graduate Civil Engineer from Aristotle University of Thessaloniki who has been working at a technical engineering office for the past two years. currently I am working on this project by myself but, if I am given funding, I shall be able to collaborate with my thesis advisor, who is an authority in this field and has a doctorate and long experience in presenting papers at conferences related to coastal engineering. I do not have experience in the application of AI in this area yet but achieving that is part of the goal. I have successfully fulfilled my thesis, which was defended before a three-member committee at my university and further reviewed by the Technical Chamber of Greece (TEE).

How much money have you raised in the last 12 months, and from where?

This is the first time I am seeking funding, and i have not received any funding for this specific project to date