Isla Duporge

Intercepting an evasive, maneuvering target is among the most computationally demanding tasks a predator performs: in the terminal phase of a chase, it must continuously convert sensory information about the target into steering and speed commands, subject to its own biomechanical limits. How terrestrial predators solve this in real time has remained difficult to quantify. Here we combine drone videography with AI-based markerless pose estimation to reconstruct the stride-resolved kinematics of 67 lion (Panthera leo) attacks on a mechanized lure programmed to move unpredictably. Lion steering is described by a combination of proportional navigation and proportional pursuit, a mixed-guidance laws previously identified only in the aerial pursuit of Harris's hawks (Parabuteo unicinctus), while speed is regulated within a defined kinematic envelope during turns, declining at close range where the cost of overshooting is greatest. These findings reveal shared guidance principles across aerial and terrestrial pursuit, providing a quantitative framework for comparing pursuit strategies across species.

Detecting wildebeest from very-high-resolution (VHR) satellite imagery enables large-area population monitoring in a single acquisition, avoiding aircraft-induced disturbance and reducing sampling bias caused by transect-based surveys. However, wildebeest appear as extremely small objects in satellite images, and direct application of classical object detectors (e.g., YOLO-style detectors) often yields poor performance. In particular, high-density aggregation areas suffer from severe missed detections due to scale mismatch and limitations in post-processing for densely packed small objects.To address these challenges, we develop a targeted detection solution that integrates (1) an adaptive sliding-window strategy to better capture local context under varying density conditions, (2) resolution–detector adaptation to mitigate scale mismatch between object size and detector design, and (3) improved post-processing modules, including an enhanced non-maximum suppression (NMS) tailored for dense small-object scenarios. We evaluate the proposed framework using WorldView-2 and WorldView-3 imagery over the Serengeti acquired in 2022 and 2023. The overall F1-score improves from 0.727 to 0.770 in 2022 and from 0.682 to 0.756 in 2023. Notably, in high-density areas in 2022, the F1-score increases from 0.330 to 0.821, demonstrating that our approach effectively reduces missed detections in dense small-object scenarios that commonly lead to substantial omissions in traditional pipelines.Beyond wildebeest monitoring, our results highlight a generalizable pathway for adapting classical detectors to dense small-object detection in VHR satellite imagery, where objects are tiny and crowded.

Recent advances in deep learning have greatly enhanced the accuracy and scalability of animal re-identification by automating the extraction of subtle distinguishing features from images and videos. This enables large-scale, non-invasive monitoring of animal populations. This article proposes a segmentation pipeline and a re-identification model to identify animals without ground-truth IDs. The segmentation pipeline isolates animals from the background using bounding boxes and leverages the DINOv2 and Segment Anything Model 2 (SAM2) foundation models. For re-identification, Recurrence over Video Frames (RoVF) is introduced, a novel approach that employs a recurrent component based on the Perceiver transformer atop a DINOv2 image model, iteratively refining embeddings from video frames. The proposed methods are evaluated on video datasets of meerkats and polar bears (PolarBearVidID). The proposed segmentation model achieved high accuracy (94.36% and 97.26%) and IoU (73.14% and 92.77%) for meerkats and polar bears, respectively. RoVF outperformed frame- and video-based re-identification baselines, achieving a top-1 accuracy of 46.5% and 55% on masked test sets for meerkats and polar bears, respectively, as well as higher top-3 accuracy. These results highlight the potential of the proposed approach to reduce annotation burdens in future individual-based ecological studies. The code is available at https://github.com/Strong-AI-Lab/RoVF-Meerkat-Reidentification .

Rhinoceroses face severe threats from poaching, habitat fragmentation, and ongoing habitat degradation. Monitoring rhinoceros across the vast, often inaccessible landscapes they inhabit is challenging. In this study, we assess the feasibility of detecting white rhinoceroses using very high-resolution (33-36 cm) satellite imagery acquired over the world’s largest private rhinoceros reserve in South Africa using a YOLO-based object detection model (YOLOv12x). We test whether synthetic imagery enhances model performance, whether rhinoceroses can be reliably distinguished from elephants in satellite imagery, and whether synthetically generated rhinoceroses are visually distinguishable from real ones by human annotators. We achieve an average precision (AP) of 0.65 in detection accuracy with synthetic augmentation yielding a marginal improvement. This study provides a demonstration of monitoring rhinos using this approach and introduces an open-access dataset to support the development and testing of new models. The aim is to facilitate effective monitoring of rhinos across the vast landscapes they inhabit. Developing new detection techniques can strengthen conservation and recovery initiatives, including translocations, assessment of breeding program success, and evaluation of anti-poaching efforts.

The Great Wildebeest Migrationin the Serengeti-Mara ecosystem is a globally iconic wildlife phenomenon that supports the health and biodiversity of the region by supporting predator populations, regulating herbivore densities, and driving nutrient cycling. This study presents the first AI-powered satellite survey, using two deep learning-based models (U-Net and YOLOv8) to detect and count wildebeest over more than 4,000 km² across two consecutive years in August 2022 and 2023 with F1 scores reaching 0.830 (Precision: 0.832, Recall: 0.838). The satellite-based results show fewer than 600,000 individuals—approximately half the widely cited estimate of 1.3 million wildebeest, which has remained largely unchanged since the 1970s. While some variation may arise from differences in spatial and temporal coverage between survey methods, the satellite approach employs rigorously validated AI models with demonstrated accuracy. Rather than undermining previous methods, this discrepancy underscores the importance of using independent and complementary monitoring tools to refine population estimates and improve our understanding of wildebeest movement dynamics.