SPOTS-10
收藏资源简介:
SPOTS-10是由林肯大学工程与物理科学学院创建的一个用于机器学习算法评估的动物图案基准数据集。该数据集包含50,000张32×32的灰度图像,涵盖了十种动物物种的多样图案,如斑点、条纹等。数据集的创建过程包括从网络收集图像、去除非自然图像和重复图像、提取90×90的图像块并转换为32×32的灰度图像。SPOTS-10主要用于夜间图像中动物物种的图案识别,旨在解决现有方法在夜间图像中依赖颜色信息不足的问题,适用于野生动物研究、生物多样性和保护应用。
SPOTS-10 is an animal pattern benchmark dataset for machine learning algorithm evaluation, developed by the School of Engineering and Physical Sciences of the University of Lincoln. This dataset contains 50,000 32×32 grayscale images, covering diverse patterns across ten animal species, including spots, stripes and other natural markings. The dataset creation workflow includes collecting images from the web, removing unnatural and duplicate images, extracting 90×90 image patches and converting them into 32×32 grayscale images. SPOTS-10 is primarily designed for pattern-based animal species recognition in nighttime images, aiming to address the issue that existing methods have insufficient reliance on color information in nighttime scenarios, and is applicable to wildlife research, biodiversity and conservation applications.

- 1SPOTS-10: Animal Pattern Benchmark Dataset for Machine Learning Algorithms林肯大学工程与物理科学学院 · 2024年



