Bioacoustics for North Atlantic Right Whale Detection Based on Deep Learning
Bibliographic record
Abstract
Climate change, pollution, shipping lanes and other man-made factors are significantly impacting the biodiversity of our planet. Whales represent some of the most endangered animals of our time with some species, such as the North Atlantic right whales, have fewer than 400 individuals left. These large mammals play a crucial role in maintaining the health of our planet due to their various attributes, such as their carbon sequestration abilities. Detecting, classifying and tracking these whales can provide us with tools for their management. Passive acoustic monitoring allows researchers to identify whales through audio recordings based on their unique sounds. We propose detecting North Atlantic right whales using passive acoustic monitoring audio clips. Our data consists of whale upcalls collected in the Gulf of St. Lawrence and the Gulf of Maine. In our work, we compare various audio visualization transformations such as the wavelet transformation, Short-Time Fourier Transform and a custom algorithm. Using the visualized audio, our goal is to use various deep learning algorithms to detect whales. We employed multiple pre-trained convolutional neural network algorithms, including DenseNetl21, EfficientNetB7 and Xception. the wavelet transformation achieved the best overall results with a validation accuracy of 90% on most detection algorithms. Specifically, DenseNet169 combined with the wavelet transformation achieved the highest accuracy of 90.39%. Using ensemble learning on the best performing algorithms, we increased the accuracy of our approach to 91.86 %. Overall, our results demonstrate the strong potential of using audio visualization with deep learning algorithms to detect North Atlantic right whales. Further work in this field, such as exploring more algorithms, could further enhance the effectiveness of our approach.
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 | 0.000 |
| Open science | 0.000 | 0.001 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 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".