SyDog
收藏资源简介:
SyDog是由萨里大学创建的一个大型合成狗数据集,包含32,000张带有2D关键点和边界框坐标的图像。该数据集通过Unity游戏引擎生成,通过改变狗的外观、姿势、环境、光照条件和摄像机视角来增加多样性。SyDog旨在通过提供大量标注数据,改善动物姿态估计模型的性能,减少对劳动密集型图像标注的需求。该数据集适用于生物力学、神经科学、行为学、机器人学和娱乐行业等多个领域的研究。
SyDog is a large-scale synthetic dog dataset created by the University of Surrey, containing 32,000 images annotated with 2D keypoints and bounding box coordinates. Generated via the Unity game engine, it increases diversity by adjusting dogs' appearances, poses, environments, lighting conditions and camera viewpoints. SyDog aims to improve the performance of animal pose estimation models by providing massive annotated data, reducing the need for labor-intensive manual image annotation. This dataset is applicable to research across multiple fields including biomechanics, neuroscience, ethology, robotics and the entertainment industry.




