MammalNet
收藏资源简介:
MammalNet是由阿卜杜拉国王科技大学等多个机构合作创建的大型哺乳动物视频数据集,包含18,346个视频,总计539小时,涵盖17个目、69个科和173个哺乳动物类别。数据集通过科学分类法对哺乳动物进行分类,并专注于12种高级动物行为的研究,如狩猎和喂养幼崽。创建过程中,研究团队采用了半自动的众包方法进行视频收集和标注,确保数据的质量和多样性。MammalNet的应用领域包括动物行为识别和理解,旨在解决大规模生态监测中的挑战,如动物识别和行为检测。
MammalNet is a large-scale mammalian video dataset co-created by King Abdullah University of Science and Technology (KAUST) and multiple other institutions. It contains 18,346 videos totaling 539 hours of footage, covering 17 orders, 69 families, and 173 mammalian taxa. The dataset adopts scientific taxonomic methods to classify mammals, and focuses on research into 12 types of advanced animal behaviors such as hunting and feeding offspring. During the dataset's development, the research team utilized a semi-automated crowdsourcing approach for video collection and annotation to ensure the quality and diversity of the data. Applications of MammalNet span animal behavior recognition and understanding, aiming to address challenges in large-scale ecological monitoring, including animal identification and behavior detection.




