Road Surface and Wetness
Dataset

• Bird's eye view road patches from calibrated cameras
  • Large scale data acquisition on test track
• Three surface types: asphalt, cobblestone, concrete
• Accurate measurements for wetness annotation

Download: balanced dataset with equal number of images for each of the 12 categories (3 surface types; 4 wetness levels)

• three patch sizes: s,m,l
• four patch distances: 7.5m, ..., 30m

• 06-20-2022, 5:20 pm: Spotlight/Poster presentation

• Supplementary video

• Paper at Computer Vision Foundation (CVF)
• Paper at ieeexplore

Base Dataset -

Road Patches

• Patch size small: 2.56 m²
• Distance to vehicle d2v=7.5m

• Patch size medium: 7.84 m²
• Distance to vehicle d2v=7.5m

• Patch size large: 12.96 m²
• Distance to vehicle d2v=7.5m

• Patch size small: 2.56 m²
• Distance to vehicle d2v=15.0m

• Patch size medium: 7.84 m²
• Distance to vehicle d2v=15.0m

• Patch size large: 12.96 m²
• Distance to vehicle d2v=15.0m

• Patch size small: 2.56 m²
• Distance to vehicle d2v=22.5m

• Patch size medium: 7.84 m²
• Distance to vehicle d2v=22.5m

• Patch size large: 12.96 m²
• Distance to vehicle d2v=22.5m

• Patch size small: 2.56 m²
• Distance to vehicle d2v=30.0m

• Patch size medium: 7.84 m²
• Distance to vehicle d2v=30.0m

• Patch size large: 12.96 m²
• Distance to vehicle d2v=30.0m

Close-to-Distribution (CtD) Datasets -

Road Patches
for domain change evaluation [1]

• Camera mounted on truck

• Camera mounted on car

• Camera mounted on car

RoadSaW Paper:

[1] K. Cordes, C. Reinders, P. Hindricks, J. Lammers, B. Rosenhahn, H. Broszio: "RoadSaW: A Large-Scale Dataset for Camera-Based Road Surface and Wetness Estimation", CVPR Workshop on Autonomous Driving, pp. 4440-4449, 2022

RoadSaW Dataset by VISCODA GmbH is licensed under CC BY-NC-SA 4.0.