• The SkyScapes dataset

    High-quality annotations, 16 high-resolution aerial imagery
    Dataset Overview

  • The SkyScapes dataset

    70,000 object instancies, 31 semantic classes
    Dataset Overview

  • The SkyScapes dataset

    Pixel-wise semantic segmentation, multi-class edge detection, multi-class lane-marking segmentation
    Benchmark Suite

The SkyScapes dataset

SkyScapes respresents a new large-scale dataset that contains a diverse set of aerial images in aerial scenes from multiple different locations with fine-grained high-quality pixel-level annotations comprised of 70000 instance objects including lane-markings. The dataset is thus by far the most diverse dataset than similar previous attempts. Details on annotated classes and examples of our annotations are available at this webpage.

TheSkyScapes Dataset is intended for

  1. assessing the performance of vision algorithms for major tasks of semantic aerial scene understanding: pixel-level labeling
  2. supporting research that aims to exploit small and large objects for training deep neural networks.

Latest News

October 1st 2023: Airborne-Shadow (ASD) dataset becomes online

Airborne-Shadow:  Towards Fine-Grained Shadow Detection in Aerial Imagery

Upcoming: Large-scale SkyScapes and SpaceScapes

The upcoming datasets will be available via this website for research purposes.

February 1st 2020: EAGLE dataset is online

Eagle dataset is now available for download.

EAGLE:  Large-Scale Vehicle Detection Dataset in Real-World Scenarios using Aerial Imagery

October 1st 2019: Registration is open

Use the registration link to create an account.

October 1st 2019: SkyScapes dataset is online

SkyScapes is now available for download and benchmarking. […]


This SkyScapes Dataset is made freely available to academic and non-academic entities for non-commercial purposes such as academic research, teaching, scientific publications, or personal experimentation. Permission is granted to use the data given that you agree to our license terms.