METU Dataset: A Big Dataset for Benchmarking Trademark Retrieval

Tursun, Osman
Kalkan, Sinan
Trademark retrieval (TR) is the problem of retrieving similar trademarks (logos) for a query, and the main aim is to detect copyright infringements in trademarks. Since there are millions of companies worldwide, automatically retrieving similar trademarks has become an important problem, and currently, checking trademark infringements is mostly performed manually by humans. However, although there have been many attempts for automated TR, as also acknowledged in the community, the problem is largely unsolved. One of the main reasons for that is the unavailability of a publicly available comprehensive dataset that includes the various challenges of the TR problem. In this article, we propose and introduce a large dataset composed of more than 930,000 trademarks, and evaluate the existing approaches in the literature on this dataset. We show that the existing methods are far from being useful in such a challenging dataset, and we hope that the dataset can facilitate the development of better methods to make progress in the performance of trademark retrieval systems.
14th IAPR International Conference on Machine Vision Applications (MVA)


Üç boyutlu marka olarak korunan ikonik mobilya tasarımları
Oğuz, Ezgi; Korkut, Fatma (null; 2018-12-28)
Mobilya tasarımları, fikir ve sanat eserleri sahiplerinin hakları, patent, tasarım, marka ve haksız rekabet gibi birçok fikri mülkiyet koruma türüne konu olurlar. Üç boyutlu marka koruması, marka hukukunda gelişen bir alan olmasına ve dikkate değer ekonomik etkisine rağmen fikri mülkiyet ve tasarım alanyazınında yeterince irdelenmemiştir. Bu çalışmanın amacı, üç boyutlu marka olarak korunan mobilya tasarımlarının tasarım ve fikri mülkiyet biyografilerini araştırarak üç boyutlu marka koruması ile diğer ...
Iconic designs protected as three-dimensional marks: the intersection of design history and intellectual property protection history
Oğuz, Ezgi; Korkut, Fatma; Department of Industrial Design (2018)
Three-dimensional marks are mainly described as the borderline between trademark and design. The three-pointed star emblem of Mercedes-Benz vehicles, the triangular prism shape of Toblerone chocolate bar, the rounded body shape of Clipper lighter, and the ‘hobble-skirt’ contour design of Coca-Cola bottle are among the well-known examples of three-dimensional marks with iconic designs coupled with economic success stories. The aim of this study is to examine three-dimensional marks in relation to other types...
Analyzing deep features for trademark retrieval Marka Erişimi İcin Derin Özniteliklerin İncelenmesi
Aker, Cemal; Tursun, Osman; Kalkan, Sinan (2017-05-18)
The rapid rise in the amount of trademark applications and trademark infringements has led the trademark retrieval (TR) to become an important and formidable task to solve. Existing studies based on hand-crafted features show unsatisfying performance. Taking the popularization and increasing success of the deep learning methods into consideration, in this work, many well-known Convolutional Neural Network (CNN) models are applied to the TR problem. Models are tested using a large scale trademark dataset in ...
Local operator spaces, unbounded operators and multinormed C*-algebras
Dosiev, Anar (Elsevier BV, 2008-10-01)
In this paper we propose a representation theorem for local operator spaces which extends Ruan's representation theorem for operator spaces. Based upon this result, we introduce local operator systems which are locally convex versions of the operator systems and prove Stinespring theorem for local operator systems. A local operator C*-algebra is an example of a local operator system. Finally, we investigate the injectivity in both local operator space and local operator system senses, and prove locally conv...
Multisource region attention network for fine-grained object recognition in remote sensing imagery
Sümbül, Gencer; Cinbiş, Ramazan Gökberk; Aksoy, Selim (Institute of Electrical and Electronics Engineers (IEEE), 2019-07)
Fine-grained object recognition concerns the identification of the type of an object among a large number of closely related subcategories. Multisource data analysis that aims to leverage the complementary spectral, spatial, and structural information embedded in different sources is a promising direction toward solving the fine-grained recognition problem that involves low between-class variance, small training set sizes for rare classes, and class imbalance. However, the common assumption of coregistered ...
Citation Formats
O. Tursun and S. Kalkan, “METU Dataset: A Big Dataset for Benchmarking Trademark Retrieval,” presented at the 14th IAPR International Conference on Machine Vision Applications (MVA), Tokyo, JAPAN, 2015, Accessed: 00, 2020. [Online]. Available: