ABOUT ESPADA

ESPADA: Extended Synthetic and Photogrammetric Aerial-image Dataset

We present a new aerial image dataset, named ESPADA, aimed to train deep neural networks for depth image estimation from a single aerial image. Given the difficulty of creating aerial image datasets containing image pairs of chromatic images related to their depth images, simulators such as AirSim have been proposed to generate synthetic images from photorealistic scenes. The latter enables the generation of thousands of images that can be used to train and evaluate neural models.

ESPADA was publicated in RA-L IEEE and will be presented in IROS 2021.

However, we argue that synthetic photorealistic aerial image datasets can be improved by adding images generated from photogrammetric models imported into the simulator, thus enabling a less artificial generation of both, chromatic and depth images. To assess the quality of these images, we compare the performance of 4 deep neural networks whose pretrained models and code for re-training are publicly available. We also use ORB-SLAM, in its RGB-D version, to indirectly assess the estimated depth image. For that, chromatic images from 3 aerial videos and their depth images, estimated with the networks trained with ESPADA, are fed into ORB-SLAM. The estimated camera pose is compared against the trajectory retrieved from the GPS flight trajectory. Our results indicate that images generated from photogrammetric models contribute to the aerial depth image estimation.

Image distribution in ESPADA:

Photogrammetry

Synthetic images

Total

Distribution of environments in ESPADA:

Photogrammetric environments

Synthetic environments

Total

DOWNLOAD

The dataset is available in a permanent link, click the button below to get access to it (Please, fill out the form, it is used for download statistics only). Be sure to have enough memory available since the dataset contains about 30 GB in data. After access is granted, a new page will be shown with download optios. Each button links to every data type described in the paper: RGB images stands for chromatic PNG images. Depth NumPy button links to NumPy files in floating point used in ESPADA. Depth PNG grayscale images button links to 8 bits PNG file images in grayscale (each pixel value represents 1 meter in distance). The size of every image is 640 (width) and 480 (height) pixels.
Also we provide the files used in the ORB-SLAM evaluation such as videos and GPS files linked in the Aerial-Videos button.

Please fill the next form to download ESPADA.

CITATION

If you use this dataset, please use the information below for citation:

R. Lopez-Campos, J. Martinez-Carranza. "ESPADA: Extended Synthetic and Photogrammetric Aerial-image Dataset" in IEEE Robotics and Automation Letters, October, 2021. DOI: https://doi.org/10.1109/LRA.2021.3101879

BibtTex citation:

@ARTICLE{9507349,
author={Lopez-Campos, Rafael and Martinez-Carranza, Jose},
journal={IEEE Robotics and Automation Letters},
title={ESPADA: Extended Synthetic and Photogrammetric Aerial-image Dataset},
month={October},
year={2021},
volume={},
number={},
pages={1-1},
doi={10.1109/LRA.2021.3101879}}

RELATED PAPERS

A. Osuna-Coutiño, J. Martinez-Carranza. “Structure Extraction in Urbanized Aerial Images from a Single View Using a CNN-based Approach”. International Journal of Remote Sensing. March, 2020. https://doi.org/10.1080/01431161.2020.1767821

L. Pellegrin, J. Martinez-Carranza. “Towards Depth Estimation in a Single Aerial Image”. International Journal of Remote Sensing. August, 2019. https://doi.org/10.1080/01431161.2019.1681601

L. O. Rojas-Perez, R. Munguia-Silva, J. Martinez-Carranza. "Real-Time Landing Zone Detection for UAVs using Single Aerial Images". IMAV2018, 10th International Micro Air Vehicle Conference (IMAV). Melbourne, Australia. November, 2018.