Datasets

ViaSign Signboard Classification

The ViaSign Dataset (VSD) is a large-scale, open-source image classification dataset containing 78,000 manually cropped Spanish signboards from 25,000 high-resolution street panoramas. The images, resized to 64×64 pixels, are annotated across 142 classes, including regulatory, warning, informational, and commercial signs, with augmented samples to address class imbalance.

VSD provides a high-quality resource for developing and benchmarking robust, language-aware traffic sign recognition models.