Data from: Snapshot Serengeti, high-frequency annotated camera trap images of 40 mammalian species in an African savanna

Camera traps can be used to address large-scale community ecology questions by providing systematic data on an array of wide-ranging species. We deployed 225 camera traps across 1,125km2 in the centre of Serengeti National Park, Tanzania, to evaluate spatial and temporal inter-species dynamics. The cameras have operated continuously since 2010. From June 2010 to May 2013, the survey accumulated 101,315 camera-trap days and produced 1.51 million sets of pictures. Members of the general public classified the images via the citizen science website www.snapshotserengeti.org. Each image was viewed by multiple users who recorded the species present, number of individuals, behaviours observed, and presence of young. For the data described here, 28,040 registered users contributed 10.8 million classifications. We applied a simple algorithm to reduce the classifications into a final "consensus" dataset, yielding a final classification for each image and a measure of agreement among individual answers. The consensus dataset and images are a novel dataset for ecological research, providing an unparalleled opportunity to investigate multi-species dynamics among dozens of species in an intact ecosystem. The consensus dataset, raw classification dataset, and images are also a valuable resource for machine-learning and computer-vision research.相机陷阱（camera trap）可通过为多种广布物种提供系统性调查数据，助力解答大型群落生态学研究课题。本研究于坦桑尼亚塞伦盖蒂国家公园核心区域的1125平方公里范围内布设了225台相机陷阱，以解析物种间的时空动态规律。该监测阵列自2010年起持续运行，在2010年6月至2013年5月的调查周期内，累计有效相机工作日达101315天，共采集到151万组影像素材。普通公众可通过公民科学平台www.snapshotserengeti.org完成影像标注工作。每张影像均由多名用户协同标注，记录其中出现的物种、个体数量、观测到的行为以及幼崽存在状态。针对本次发布的数据集，共有28040名注册用户完成了总计1080万次标注任务。本研究通过简易算法整合多轮标注结果，生成最终的‘共识’数据集，为每张影像提供统一分类结果，并量化各标注者答案间的一致性程度。本共识数据集及配套影像为生态学研究提供了全新的宝贵资源，为探究完整自然生态系统中数十个物种种间的多维度动态关系提供了前所未有的研究契机。此外，共识数据集、原始标注数据集及配套影像同样可为机器学习与计算机视觉领域的研究提供重要支撑。