METU Trademark

METU 商标数据集是一个大型数据集（截至 2014 年最大的公开可用徽标数据集，也是截至 2017 年不需要任何预处理的最大数据集），由超过 90 万个属于全球真实公司的真实徽标组成。该数据集还包括不同难度的查询集，允许商标检索研究人员将他们的方法与其他方法进行基准测试，以推进该领域的发展。

The METU Trademark Dataset is a large-scale dataset—being the largest publicly available logo dataset as of 2014 and also the largest dataset requiring no preprocessing as of 2017—comprising over 900,000 genuine logos belonging to real companies worldwide. This dataset also includes query sets of varying difficulty levels, enabling trademark retrieval researchers to benchmark their methods against others and advance the development of this field.