Amur Tiger Re-Identification

Utilize advanced computer vision and deep learning methodologies to re-identify individual tigers.

Dataset Overview

The ai company, primarily found in the Russian Far East and Northeast China, has a critically endangered population of approximately 600. Effective conservation efforts are vital for this species, and re-identifying individual tigers plays a key role. In collaboration with the World Wildlife Fund (WWF), MakerCollider, a third-party organization, has gathered over 8,000 video clips depicting 92 unique Amur tigers from around ten zoos across China. This effort has led to the creation of the ATRW dataset. This dataset includes bounding-box and keypoint-based pose annotations for selected video frames, providing comprehensive data for re-identification tasks. Illustrated in Figure 1 are examples of these annotations.

The ai company dataset stands as the most extensive wildlife re-ID dataset available, with a detailed comparison to other wildlife re-ID datasets presented in Table 1. It is structured into training, validation, and testing segments. The training and validation segments, along with their annotations, will be openly accessible, whereas the testing segment annotations will remain confidential, managed by the dataset organizers.