Benchmarking coastal boundary datasets in deep learning applications
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Coastal areas are of high importance for both human development and Earth’s biosphere. However, coastal erosion threatens the balance between the geosphere, hydrosphere and biosphere, putting fragile ecosystems at risk. Human activities such as urbanization and climate change further exacerbate this natural phenomenon. Effective monitoring solutions are essential for mitigation strategies and for understanding land–ocean interactions. Given the vast size of coastal regions, automated tools for monitoring coastal boundaries are increasingly necessary. Artificial intelligence, particularly deep learning combined with remote sensing data, has shown promise in this domain. Currently, there is a lack of benchmarking studies for datasets relevant to this task. This study aims to fill that gap by comparing multiple remote sensing datasets for boundary extraction using deep learning. Benchmarking available datasets and models provides a foundation for standardization and future workflow integration. Seven datasets were compared and cross-tested using three popular deep learning algorithms. A novel metric based on pixel-level edge accuracy was developed and used to evaluate model performance. The results demonstrate the ability of deep learning algorithms to generalize efficiently across multiple datasets. The SNOWED dataset, in particular, achieved highly promising results with strong cross-dataset F1-scores, accurate boundary delineation and robust generalization. These findings highlight the potential for a resolution-agnostic and reliable framework for coastal boundary extraction using optical satellite imagery.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.002 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it