MachineLearningVisionRG / GRATS

GRATS (GReek Adversarial Traffic Signs)

This repository contains a dataset that proposed in the paper "Traffic Signs Recognition Robustness in Autonomous Vehicles under Physical Adversarial Attacks" and consists of clean and 'dirty' images belonging to five classes, no_parking, no_right_turn, no_left_turn, no_entry and stop.

Dataset Structure

The dataset follows the following structure:

├── Dataset
    ├── clear_road_signs
        ├── train
            ├── no_righ_turn
            ├── no_parking
            ...
        ├── test
            ├── no_righ_turn
            ├── no_parking
            ...
    ├── dirty_road_signs
        ├── train
            ├── no_righ_turn
            ├── no_parking
            ...
        ├── test
            ├── no_righ_turn
            ├── no_parking
            ...

Citation

If you use our dataset in a scientific publication, please use the following citation:

Apostolidis, K.D., Gkouvrikos, E.V., Vrochidou, E., Papakostas, G.A. (2023). Traffic Sign Recognition Robustness in Autonomous
Vehicles Under Physical Adversarial Attacks. In: Daimi, K., Alsadoon, A., Coelho, L. (eds) Cutting Edge Applications of
Computational Intelligence Tools and Techniques. Studies in Computational Intelligence, vol 1118. Springer,
Cham. https://doi.org/10.1007/978-3-031-44127-1_13