METU Trademark Dataset
收藏arXiv2017-10-14 更新2024-06-21 收录
下载链接：
http://kovan.ceng.metu.edu.tr/LogoDataset/
下载链接
链接失效反馈官方服务：
资源简介：
METU Trademark Dataset是由中东技术大学KOVAN研究实验室创建的大型商标数据集，包含约923,343个商标图像，涵盖约410,000家公司的商标。该数据集旨在为商标检索问题提供一个大规模的基准，通过结合多种视觉特征和深度学习方法，评估不同商标检索系统的性能。数据集中的商标图像经过清理和优化，以适应学术研究的需求。该数据集的应用领域主要集中在商标检索和相似商标的识别，旨在解决商标侵权和品牌保护等实际问题。
METU Trademark Dataset is a large-scale trademark dataset created by the KOVAN Research Lab at Middle East Technical University. It contains approximately 923,343 trademark images covering trademarks from around 410,000 companies. This dataset aims to provide a large-scale benchmark for trademark retrieval tasks, and evaluate the performance of different trademark retrieval systems by combining multiple visual features and deep learning methods. The trademark images in the dataset have been cleaned and optimized to meet the requirements of academic research. Its main application areas focus on trademark retrieval and similar trademark recognition, aiming to solve practical issues such as trademark infringement and brand protection.
提供机构：
中东技术大学
创建时间：
2017-01-20
