GC3558: An open-source annotated dataset of Ghana currency images for classification modeling

• Published in: Data in brief Vol. 45; p. 108616
• Main Authors: Adu, Kwabena, Mensah, Patrick Kwabena, Ayidzoe, Mighty Abra, Appiah, Obed, Quayson, Ebenezer, Ninfaakang, Christopher Bombie, Opoku, Michael
• Format: Journal Article
• Language: English
• Published: Elsevier Inc 01-12-2022; Elsevier
• Subjects: Banknote recognition; Classification; Currency detection; Data; Dataset; Deep learning; Deep learning; Banknote recognition; Dataset; Classification; Currency detection
• ISSN: 2352-3409; 2352-3409
• DOI: 10.1016/j.dib.2022.108616

The field of deep learning has led to remarkable advancements in many areas, including banking. Identifying currency denomination type and model is challenging due to intraclass variation and different illumination conditions. Although, in this domain, many datasets regarding currency denomination type and model, e.g., Indian Currency, Thai Currency, Chinese Currency, U.K. currency, etc., have already been experimented with by different researchers. More datasets are needed from a variety of currencies, especially Ghana currency (cedi). This article presents the Ghana Currency image dataset (GC3558) of 3558 color images in 13 classes created from a high-resolution camera. The dataset is comprised of only genuine currency. The class consists of coin and paper notes: 10 pesewas coin, 20 pesewas coin, 50 pesewas coin, 1 cedi coin, 2 cedis coin, 1 cedi note, 2 cedis note, 5 cedis note, 10 cedis note, 20 cedis note, 50 cedis note, 100 cedis note and 200 cedis note. All images are de-identified, validated, and freely available for download to A.I. researchers. The dataset will help researchers evaluate their machine learning models on real-world data.