Perch 2.0 is a supervised bioacoustics foundation model pretrained on 14,597 species, including birds, mammals, amphibians, and insects, and has state-of-the-art performance on multiple benchmarks. Given that Perch 2.0 includes almost no marine mammal audio or classes in the training data, we evaluate Perch 2.0 performance on marine mammal and underwater audio tasks through few-shot transfer learning. We perform linear probing with the embeddings generated from this foundation model and compare performance to other pretrained bioacoustics models. In particular, we compare Perch 2.0 with previous multispecies whale, Perch 1.0, SurfPerch, AVES-bio, BirdAVES, and Birdnet V2.3 models, which have open-source tools for transfer-learning and agile modeling. We show that the embeddings from the Perch 2.0 model have consistently high performance for few-shot transfer learning, generally outperforming alternative embedding models on the majority of tasks, and thus is recommended when developing new linear classifiers for marine mammal classification with few labeled examples.

Automated detection and classification of marine mammals vocalizations is critical for conservation and management efforts but is hindered by limited annotated datasets and the acoustic complexity of real-world marine environments. Data augmentation has proven to be an effective strategy to address this limitation by increasing dataset diversity and improving model generalization without requiring additional field data. However, most augmentation techniques used to date rely on effective but relatively simple transformations, leaving open the question of whether deep generative models can provide additional benefits. In this study, we evaluate the potential of deep generative for data augmentation in marine mammal call detection including: Variational Autoencoders, Generative Adversarial Networks, and Denoising Diffusion Probabilistic Models. Using Southern Resident Killer Whale (Orcinus orca) vocalizations from two long-term hydrophone deployments in the Salish Sea, we compare these approaches against traditional augmentation methods such as time-shifting and vocalization masking. While all generative approaches improved classification performance relative to the baseline, diffusion-based augmentation yielded the highest recall (0.87) and overall F1-score (0.75). A hybrid strategy combining generative-based synthesis with traditional methods achieved the best overall performance with an F1-score of 0.81. We hope this study encourages further exploration of deep generative models as complementary augmentation strategies to advance acoustic monitoring of threatened marine mammal populations.

Breakthroughs, discoveries, and DIY tips sent every weekday. Orca whales (Orcinus orca) are among the most fearsome apex predators in the ocean. One well-studied group of orcas living between Washington State and British Columbia now has a new skill to add to its repertoire–tool manufacturing and use. The whales create tools from kelp and appear to use them to help groom one another. The findings are detailed in a study published June 23 in the Cell Press journal Current Biology.

For nearly 100 years, scientists have been perplexed by the famous Nazca geoglyphs – ancient patterns in the soil of the Nazca Desert in southern Peru. Now, with the help of AI, researchers have discovered another 303 drawings – and they're possibly the most bizarre yet. Among them are alien-like figures, killer whales holding knives, cats, camels and a figure that looks like Pixar's Wall-E robot. Photos show some of the new discoveries, with lines manually added on the images to emphasize the original lines, which have faded due to erosion. The mysterious Nazca glyphs may date back to 400 BC, but scientists are still unsure what their exact purpose was, if any.

Hundreds of ancient drawings depicting decapitated human heads and domesticated llamas have been discovered in the Peruvian desert with the help of artificial intelligence. Archaeologists have previously linked these creations to the people of the Nazca culture, who started etching such images, called geoglyphs, into the ground around 2000 years ago. These geoglyphs are smaller and older than the Nazca lines and other figures found to date, which portray huge geometric shapes stretching several kilometres or wild animals about 90 metres long on average. The newly discovered images typically depict humanoid figures and domesticated animals around 9 metres long. Some even hint at human sacrifice, portraying decapitated heads and killer whales armed with blades.

Orcinus orca (killer whales) exhibit complex calls. They last about a second. In a call, an orca typically uses multiple frequencies simultaneously, varies the frequencies, and varies their volumes. Behavior data is hard to obtain because orcas live under water and travel quickly. Sound data is relatively easy to capture. As a science goal, we would like to know whether orca vocalizations constitute a semantic language. We do this by studying whether machine learning can predict behavior from vocalizations. Such prediction would also help scientific research and safety applications because one would like to predict behavior while only having to capture sound. A significant challenge in this process is lack of labeled data. We work with recent recordings of McMurdo Sound orcas [Wellard et al. 2020] where each recording is labeled with the behaviors observed during the recording. This yields a dataset where sound segments - continuous vocalizations that can be thought of as call sequences or more general structures - within the recordings are labeled with superfluous behaviors. Despite that, with a careful combination of recent machine learning techniques, we achieve 96.4% classification accuracy. This suggests that orcas do use a semantic language. It is also promising for research and applications.

Researchers have discovered another 168 geoglyphs made in the soil of Peru's Nazca Desert, known as the Nazca lines. The newly-discovered drawings – identified by a team at Yamagata University in Japan – depict humans, camelids, birds, killer whales, felines and snakes. One of the human drawings looks like Homer Simpson, with big cartoon eyes and a patch of what looks like stubble around the mouth. These 168 newly-found geoglyphs are thought to date between 100 BC and AD 300, according to experts, but other Nazca lines may go back even further to 400 BC. The Nazca lines are a group of geoglyphs made in the soil of the Nazca Desert in southern Peru.

Conventional saliency maps highlight input features to which neural network predictions are highly sensitive. We take a different approach to saliency, in which we identify and analyze the network parameters, rather than inputs, which are responsible for erroneous decisions. We first verify that identified salient parameters are indeed responsible for misclassification by showing that turning these parameters off improves predictions on the associated samples, more than turning off the same number of random or least salient parameters. We further validate the link between salient parameters and network misclassification errors by observing that fine-tuning a small number of the most salient parameters on a single sample results in error correction on other samples which were misclassified for similar reasons - nearest neighbors in the saliency space. After validating our parameter-space saliency maps, we demonstrate that samples which cause similar parameters to malfunction are semantically similar. Further, we introduce an input-space saliency counterpart which reveals how image features cause specific network components to malfunction.

Killer whales have been caught on camera hunting and eating Great White Sharks for the first time, in an hour-long feeding frenzy. The extraordinary scenes were shot by both helicopter and drone pilots off the coast of South Africa, providing the first direct evidence of orcas preying on sharks. They reveal that the killer whales were attacking at least four sharks for about an hour, and that this unusual predatory behaviour might be spreading in the species. While a short drone video of the attack was released in June, a paper has been released this month analysing the clip plus all the footage taken from the helicopter. Scientists from the Marine Dynamics Academy studied the videos, and analysed drone and cage dive boat survey data before and after these predation events.