Penn Aviary Dataset

Penn Aviary Dataset是由宾夕法尼亚大学创建的一个专注于鸟类三维重建的数据集。该数据集包含15只社会性鸟类在户外鸟舍中的行为记录，涵盖了三个月的交配季节。数据集通过八个同步摄像头捕捉，记录了鸟类的姿态、方向和位置的广泛变化，以及不同光照条件和季节的影响。创建过程中，研究人员对1000个实例进行了轮廓和关键点标注，并匹配了这些标注跨视角。该数据集主要用于研究社会性鸟类的行为，旨在解决从单一视角准确估计鸟类姿态和形状的问题，对于理解鸟类如何在复杂环境中导航、决策以及社会互动的形成具有重要意义。

The Penn Aviary Dataset is a specialized dataset focused on 3D reconstruction of birds, developed by the University of Pennsylvania. It contains behavioral recordings of 15 social birds in an outdoor aviary, spanning a three-month mating season. Captured by eight synchronized cameras, the dataset records extensive variations in the birds' postures, orientations and positions, as well as the effects of different lighting conditions and seasonal changes. During its creation, researchers performed contour and keypoint annotations for 1,000 instances, and matched these annotations across different viewpoints. This dataset is mainly used for researching the behavior of social birds, aiming to solve the problem of accurately estimating bird posture and shape from a single viewpoint, and is of great significance for understanding how birds navigate, make decisions and form social interactions in complex environments.