Individual tree crown delineation in high resolution aerial RGB imagery using StarDist-based model
Bibliographic record
Abstract
The availability of high spatial resolution remote sensing imagery has facilitated forestry attribute estimation at the individual tree level. However, producing accurate tree crown delineations for practical applications remains challenging, particularly in mixed forests with overlapping tree crowns. In this study, we propose an individual tree crown delineation method leveraging the StarDist model to improve the delineation accuracy in mixed forests. The StarDist model captures tree crown shapes uniquely through star-convex polygons, which are predicted by the U-Net architecture. The final tree crowns are determined by applying non-maximum suppression (NMS) to all identified star-convex polygons. Performance evaluation on two mixed forest areas reveals a delineation accuracy exceeding 92%, notably outperforming the widely used deep learning model MASK R-CNN by over 6%. In terms of tree crown areas estimation, the R 2 for both testing areas is higher than 0.85 for both testing areas. Moreover, the evaluations on precision, recall, and F1-score demonstrate that the proposed model can generate tree crowns fitting well with the true crowns. This study marks the first utilization of the StarDist model for tree crown delineation in mixed forests. Our findings demonstrate the effectiveness of the StarDist model for accurately delineating individual tree crowns, thereby advancing the field of forestry research. • We mark the first utilization of the StarDist model for the tree crown delineation task. • The star-convex polygon is used to describe the shape of tree crowns. • The algorithm was tested in two mix-forest areas. • A delineation accuracy exceeding 92% is achieved with a small training set. • The StarDist model notably surpasses the commonly-used deep learning model MASK R-CNN.
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 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 | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| 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 itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, 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".