MammalNet

MammalNet视频数据集围绕涵盖 17 个目、69 个科和 173 个哺乳动物类别的生物哺乳动物分类学构建，并包括 12 种常见的高级哺乳动物行为（例如狩猎、梳理行为）。
MammalNet 能够单独或联合研究动物和行为识别。它还有助于研究具有挑战性的构图场景，测试模型的零样本和低样本迁移能力。发表于CVPR2023，https://mammal-net.github.io/

The MammalNet video dataset is built upon the biological mammalian taxonomy covering 17 orders, 69 families, and 173 mammalian species, and includes 12 common advanced mammalian behaviors (e.g., hunting, grooming).
MammalNet enables individual or joint research on animal and behavior recognition. It also facilitates the study of challenging compositional scenes and tests the zero-shot and few-shot transfer capabilities of models. Published at CVPR 2023, https://mammal-net.github.io/