Using Deep Learning to Automate Orangutan Nest Detections on Aerial Images Collected With Drones
Bibliographic record
Abstract
Traditional orangutan distribution and density monitoring requires costly line transect methods on the ground to detect their nests. Recently researchers have started to use unoccupied aerial vehicles, hereafter referred to as drones, to collect such data faster. However, manually inspecting the images acquired by the drone is time-consuming and hence costly. This study explored a deep learning method for the automated detection of orangutan nests in drone-captured aerial images, which can significantly improve the efficiency of orangutan monitoring efforts. The YOLO v10 model was trained using 868 images containing 1568 annotated orangutan nests collected from sites in Sabah, Malaysia, and Sumatra, Indonesia. Images were captured using multirotor and fixed-wing drones at varying altitudes. The model was trained using a transfer learning approach and achieved a mean Average Precision (mAP) of 0.831. The model was subsequently tested on two independent data sets with results showing a precision of 0.98 and recall of 0.88 for a multirotor drone and precision of 0.98 and a recall of 0.71 for a fixed-wing drone which has the benefit of being able to have longer duration flights. The high precision values indicate the model's accuracy in identifying true nest locations, while the recall values demonstrate its ability to detect a significant portion of the nests present in the images. The study highlights how using drones for data collection can reduce survey times compared to ground surveys, and the automation of nest detection further enhances the efficiency of drone surveys. However, the model's recall, especially for fixed-wing drone data, could be improved to ensure accurate population trend analyses. Further research should focus on expanding training data sets and refining models to account for different camera systems and environmental conditions.
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 | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.001 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 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".