Published: 17-06-2020| Version 1 | DOI: 10.17632/3bmmnfb4bp.1
Siniša Šegvić


This is a semantic segmentation dataset collected at University of Zagreb, Faculty of Electrical Engineering and Computing. A more complete dataset description is available at This dataset contains groundtruth semantic segmentations for 445 hand-picked images from the KITTI dataset. We start from 146 images annotated by German Ros from UAB Barcelona, improve their annotation accuracy and contribute another 299 images. The annotations feature high quality pixel-level polygonal approximations into 11 semantic classes: building, vegetation, sky, road, fence, pole, sidewalk, sign, car, pedestrian, bicyclist. Dataset was designed by Ivan Krešo and Siniša Šegvić. Images were selected by Ivan Krešo. Annotation was performed by Ivan Borko, Matija Folnović, Petra Marče, Nikola Munđer and Dino Pačandi. The annotation tool was designed by Ivan Fabijanić.


Steps to reproduce

If you find our dataset useful in your research, please cite the following paper: Ivan Krešo, Denis Čaušević Josip Krapac and Siniša Šegvić. Convolutional scale invariance for semantic segmentation. GCPR 2016.