"PL-Coins: A YOLO-Ready Polish Coin Dataset"

Read full text
Authors: Marcin Judt, Konrad Kamiński, Aleksandra Świetlicka
Cited by: 1

Abstract

"This dataset was developed in order to train YOLO-based object detection models, specifically for integration into mobile applications. The collection contains 1539 high-resolution (around 3000x4000 pixels or more, depending on the camera used) images of Polish coins, captured under diverse lighting conditions and backgrounds to ensure high model generalization. By including variations in coin wear, orientation, and overlapping positions, the data addresses common challenges in physical currency identification. Objects in the dataset were categorized into 13 unique classes, featuring specific distinctions for 5 PLN, 2 PLN, silver and bronze obverses alongside nine standard denominations. All images are accompanied by YOLO-compliant label files, providing bounding box coordinates for each object. This resource is provided for noncommercial research and educational purposes only."

Publication details

Journal / source
IEEE DataPort
Publication date
24 March 2026
Document type
Dataset