Cows2021
收藏资源简介:
Cows2021数据集是由布里斯托大学创建的,旨在通过视频识别个体荷斯坦-弗里斯兰牛。该数据集包含10,402张RGB图像和301个视频,捕捉了牛群中独特的黑白斑纹图案。数据集的创建过程涉及使用高可靠性的无ID牛检测器和跟踪技术,以形成标准化和增强的个体跟踪序列。Cows2021数据集主要应用于加速视觉牛识别系统的标注工作,通过自监督学习减少手动标注的负担,从而在生物识别动物(重新)识别领域推动研究进展。
The Cows2021 dataset was developed by the University of Bristol, with the objective of recognizing individual Holstein-Friesian cows via video. This dataset contains 10,402 RGB images and 301 videos, capturing the unique black-and-white stripe patterns of the cattle herd. The creation of the Cows2021 dataset employed highly reliable ID-free cattle detectors and tracking technologies to form standardized and enhanced individual tracking sequences. The Cows2021 dataset is primarily used to accelerate the annotation work of visual cow recognition systems, reduce the manual annotation burden through self-supervised learning, and thereby advance research in the field of biometric animal (re-)recognition.

- 1Towards Self-Supervision for Video Identification of Individual Holstein-Friesian Cattle: The Cows2021 Dataset布里斯托大学 · 2021年



