BuckTales : A multi-UAV dataset for multi-object tracking and re-identification of wild antelopes

TL;DR

BuckTales is introduced, the first large-scale UAV dataset designed to solve multi-object tracking (MOT) and re-identification (Re-ID) problem in wild animals, specifically the mating behaviour (or lekking) of blackbuck antelopes.

Abstract

Understanding animal behaviour is central to predicting, understanding, and mitigating impacts of natural and anthropogenic changes on animal populations and ecosystems. However, the challenges of acquiring and processing long-term, ecologically relevant data in wild settings have constrained the scope of behavioural research. The increasing availability of Unmanned Aerial Vehicles (UAVs), coupled with advances in machine learning, has opened new opportunities for wildlife monitoring using aerial tracking. However, limited availability of datasets with wild animals in natural habitats has hindered progress in automated computer vision solutions for long-term animal tracking. Here we introduce BuckTales, the first large-scale UAV dataset designed to solve multi-object tracking (MOT) and re-identification (Re-ID) problem in wild animals, specifically the mating behaviour (or lekking) of blackbuck antelopes. Collected in collaboration with biologists, the MOT dataset includes over 1.2 million annotations including 680 tracks across 12 high-resolution (5.4K) videos, each averaging 66 seconds and featuring 30 to 130 individuals. The Re-ID dataset includes 730 individuals captured with two UAVs simultaneously. The dataset is designed to drive scalable, long-term animal behaviour tracking using multiple camera sensors. By providing baseline performance with two detectors, and benchmarking several state-of-the-art tracking methods, our dataset reflects the real-world challenges of tracking wild animals in socially and ecologically relevant contexts. In making these data widely available, we hope to catalyze progress in MOT and Re-ID for wild animals, fostering insights into animal behaviour, conservation efforts, and ecosystem dynamics through automated, long-term monitoring.

Figures (4)

• Figure 1: A schematic of the data collection strategy and dataset details. The image in the top displays top-down view of a blackbuck lek from a single drone with male territories marked. The close-ups on the right show an example male and female. The bottom figure is a simplified data collection scheme, shown here with two drones (note that the actual collection scheme involved three drones). Three types of annotations are made available with the manuscript: object detection, multi-object tracking and re-identification.
• Figure 2: Sample images from the Re-ID dataset. The images in each column demonstrates the variation in appearance of an individual when captured from two different drones.
• Figure 3: A cursory movement analysis highlighting the movement and behavioural diversity captured by our dataset. A,B) show diversity in individual specific behaviours while C–G) shows the same in pairs of interacting individuals.
• Figure 4: The figure highlighs examples of some limitations of the existing dataset and the state of the art. Images on the left- who that bounding box detection does not always fit perfectly. Shadows and difficult postures can produce missing detections or create confusion with identification of class category. Proximity of the animals also contributes to error in tracking due to missing detection, merged bounding boxes or switching in IDs.