Datasets
Open Access
WRIVA Public Data
- Citation Author(s):
- Submitted by:
- Myron Brown
- Last updated:
- Wed, 12/18/2024 - 08:18
- DOI:
- 10.21227/cjk5-gf33
- Data Format:
- License:
206 Views- Categories:
- Keywords:
Abstract
The IARPA WRIVA program aims to develop software systems that can create photorealistic, navigable 3D site models using a highly limited corpus of imagery, to include ground level imagery, surveillance height imagery, airborne altitude imagery, and satellite imagery. Additionally, where imagery lacks metadata indicating geolocation, information about camera parameters, or is corrupted by artifacts, WRIVA seeks to detect and correct these factors to incorporate the imagery in site-modelling and other downstream image processing and analysis algorithms. The Johns Hopkins University Applied Physics Laboratory (JHU/APL) and the MIT Lincoln Laboratory have collected, calibrated, and curated data to support the WRIVA program. JHU/APL and the MITRE Corporation have collaborated to develop challenge datasets designed to evaluate WRIVA systems against a broad range of program objectives. IARPA is making some of this data publicly available to enable reproducible public research.
WRIVA Challenge Datasets (11/2024)
Datasets were constructed to explore sensitivity of camera calibration and novel view synthesis algorithm performance to various real-world challenges. Datasets are provided for the following challenge themes:
- Image density
- Camera models
- Varying altitudes
- Reconstructed area
- Image resolution
- PTZ detection
A zip file is provided for each theme. Each dataset folder is labeled t<xx>_v<xx>_s<xx>_r<xx>_<theme>_<details>.
- t<xx>: theme number
- v<xx>: challenge vector number for the given theme
- s<xx>: step number in the challenge vector, with each step varying the challenge level
- r<xx>: revision number for the dataset; note that only one revision is provided
- <theme>: theme name
- <details>: label indicating test site and any other identifying information
Each dataset includes input and reference folders, each with images and JSON metadata. Additional details regarding metadata schema, specific challenge dataset settings, and performance evaluation will be provided.
ULTRRA Versions of WRIVA Datasets (In work)
WRIVA datasets may also be converted to the simpler format used for the ULTRRA Challenge (https://ieee-dataport.org/competitions/ultrra-challenge-2025). This allows researchers to use any ULTRRA solution to work with our data directly.
WRIVA Full Site Data (In work)
In addition to providing pre-curated challenge datasets, we will release full datasets for some test sites and code for crafting new challenge datasets.
Acknowledgements
This work was supported by Intelligence Advanced Research Projects Activity (IARPA) contract numbers 2020-20081800401 and 2020-20090800401-017. Disclaimer: The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsements, either expressed or implied, of IARPA or the U.S. Government.
Dataset Files
wriva-challenge-datasets-varying-altitudes-1.zip (27.62 GB)- wriva-challenge-datasets-image-density-1.zip (12.94 GB)
- wriva-challenge-datasets-ptz-detection-1.zip (11.89 GB)
- wriva-challenge-datasets-artifact-injection-1.zip (11.60 GB)
- wriva-challenge-datasets-image-resolution-1.zip (7.05 GB)
- wriva-challenge-datasets-reconstructed-area-1.zip (5.47 GB)
- wriva-challenge-datasets-camera-models-1.zip (5.08 GB)
Open Access dataset files are accessible to all logged in users. Don't have a login? Create a free IEEE account. IEEE Membership is not required.
