Dataset

This is the aerial-multi-spectral-thermal dataset involved in the paper PETIT-GAN, which was published in WACV2024.
For more details about the research, please visit the project's website.

Each of the files below is a .zip file holding the entirety of the valid aerial frames acquired in that wavelength, as described in the research paper and supplementary material.

• Panchromatic
• 9000nm
• 11000nm
• 12000nm

Each frame is an independent numpy .npz file with the following fields:

• image: a two-dimensional 256 x 336 Numpy array containing the radiometric intensities acquired by the Flir TAU2 camera (with 14-bit precision).
• time: the date and time of the acquired image in GMT.
• lat: the latitude of the plane at the time of acquisition.
• long: the longitude of the plane at the time of acquisition.
• alt: the altitude of the plane above sea-level at the time of acquisition (in meters).
• fpa: the intrinsic temperature of the camera as measured by the focal-plane array thermal sensor (in Celsius).

@InProceedings{Berman_2024_WACV,
  author    = {Berman, Omri and Oz, Navot and Mendlovic, David and Sochen, Nir and Cohen, Yafit and Klapp, Iftach},
  title     = {PETIT-GAN: Physically Enhanced Thermal Image-Translating Generative Adversarial Network},
  booktitle = {Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)},
  month     = {January},
  year      = {2024},
  pages     = {1618-1627}
}