Building Detection and Outlining in Multi-Modal Remote Sensor Data: A Stratified Approach
Bibliographic record
Abstract
Exploitation of aerial sensor data for vectorized representation of urban structures is of great importance for urban modeling. Even if elevation data is available, extracting complex and irregular building shapes would become challenging. In this paper, we present a stratified approach for the extraction of building outlines consisting of two steps: Classification and vectorization. For classification, we use training data transfer to evaluate a dataset with no or little labeled data. Since the available reference data is also flawed, we use a self-developed interactive tool to adjust and improve the building shape before contrasting it with the classification results. Initial building polygons are slightly generalized and refined considering building shape characteristics. Hereby, Least-Squares adjustment is implemented to solve the best-fit problem of building edges with the input data by applying the Gauss-Helmert and Gauss-Markov Models. On the raster level, the resulting polygons achieved an accuracy of over 99% in the Potsdam dataset and almost 98% in the Munich dataset while on the vector level, the median building-wise deviations lie in sub-meter range. Although there were few mis-detections and, sometimes, considerable peak deviations for the Hausdorff distance, the compelling qualitative results attest to the robustness and validity of the proposed procedure.
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.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.003 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.003 | 0.002 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.002 | 0.001 |
| Open science | 0.001 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
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 itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".