SkyScenes: A Synthetic Dataset for Aerial Scene Understanding

Sahil Khose*, Anisha Pal*, Aayushi Agarwal*, Deepanshi*, Judy Hoffman, Prithvijit Chattopadhyay

📣 Announcements

SkyScenes has been accepted at ECCV 2024 !

🚀 Release Update

• Code under maintanence, will be released soon!

💻 Installation and Generation

• Step 0: Install docker: https://docs.docker.com/engine/install/ubuntu/

  Check out CARLA's documentation on how to setup docker for further details.

• Step 1: Setup docker

  sudo systemctl start docker
  sudo docker pull carlasim/carla:0.9.14
  sudo docker run --privileged --gpus all --net=host -v /tmp/.X11-unix:/tmp/.X11-unix:rw carlasim/carla:0.9.14 /bin/bash ./CarlaUE4.sh -RenderOffScreen

• Step 2: Open a new terminal

  To get [containerName]:

  sudo docker ps # under NAMES

  sudo docker cp [containerName]:/home/carla/PythonAPI ./

• Step 3: Data Generation

  Update the DIR to store the data along with the various height, pitch, town, weather variations inside this script.

  python3 generate_variations.py

🔧 Troubleshooting

• If you have already started a docker container and you want to gracefully stop it to re-run the commands for generation:

  sudo docker ps
  sudo docker stop [NAME or ID]

• Note: Every script should have the following snippet at the beginning before importing carla

  import sys
  sys.path.append('PythonAPI/carla/dist/carla-0.9.14-py3.7-linux-x86_64.egg')
  import carla
  

  We already have these paths in our generation scripts. You might have to change linux to windows/mac according to your system

• Generating images

  We have already generated the inital datapoints using this and saved the metaData under ./meta_data. The humanspawn() algorithm is located in this script.

  python3 manualSpawning.py

BibTex

If you find our work useful please star ⭐️ our repo and cite 📄 our paper. Thanks for your support!

@misc{khose2023skyscenes,
      title={SkyScenes: A Synthetic Dataset for Aerial Scene Understanding}, 
      author={Sahil Khose and Anisha Pal and Aayushi Agarwal and Deepanshi and Judy Hoffman and Prithvijit Chattopadhyay},
      year={2023},
      eprint={2312.06719},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}