PGTLP: A Dataset for Tunisian License Plate Detection and Recognition

Amir Ismail (1) , Mehri Maroua (1) , Anis Sahbani (2) , Najoua Essoukri Ben Amara (1)

1 LATIS - Laboratory of Advanced Technology and Intelligent Systems
2 ISIR - Institut des Systèmes Intelligents et de Robotique

The whole point throughout this work is a publicly available and annotated images dataset for vehicle license plate (LP) detection and recognition. This contribution is driving by the fact that no Tunisian LP dataset has been provided up to this point in time. Pearl Guard Tunisian LP (PGTLP) dataset is fully annotated database that contains presently up to 3, 000 images captured by high definition quality camera mounted on a security mobile robot, called Pearl Guard and made accessible to the Academia. For benchmarking and ranking supply, we propose a unified, real-time (LP) detection module based on YOLOv4-tiny detector. We also provide baseline results for vehicle LP detection using the PGTLP dataset.

Mots clés: Vehicle license plates detection recognition dataset benchmark; Vehicle license plates; detection; recognition; dataset; benchmark

Domaines: Sciences de l'ingénieur [physics]

Amir Ismail, Mehri Maroua, Anis Sahbani, Najoua Essoukri Ben Amara. PGTLP: A Dataset for Tunisian License Plate Detection and Recognition. 18th International Multi-Conference on Systems, Signals & Devices (SSD),, Mar 2021, Monastir, Tunisia. pp.661--666. ⟨hal-03910186⟩