Multilabel classification of peatland plant species from high-resolution drone images
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
Biodiversity monitoring programs are essential for detecting changes in species distributions and correlating these changes with biotic and abiotic factors. This information is crucial for identifying early problems before they become too difficult to address and for implementing effective management strategies. Traditionally, biodiversity monitoring for small plant species has relied on the quadrat method, which requires botanists to identify species in the field. While this method has its advantages, it is limited by the availability of botanists, restricting the scale of monitoring programs. In this study, we explored the potential of using high-resolution photos and artificial intelligence to estimate small plant species cover in peatlands, thereby reducing the need for field-based species identification by botanists. Our approach involves dividing quadrat images into smaller tiles, applying a multi-label classification model to each tile, and calculating species cover based on the identified tiles. Data were collected from 32 sites across Quebec, and images were annotated for five common species: Chamaedaphne calyculata , Kalmia angustifolia , Andromeda polifolia , Rhododendron groenlandicum , and Larix laricina . Our model achieved a global F1 score of 71.68 %, with the highest-performing species ( Larix laricina ) reaching 87.17 %. Although some species showed lower performance, the estimated species cover by our model in a whole quadrat was comparable to traditional methods. Our results demonstrate that this method offers significant advantages for monitoring broad changes in vegetation.
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 | 0.000 | 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) | 0.001 | 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