Model Parking Lot RGB DoFP Dataset

The goal of this dataset is to generate visible spectro-polarimetric image data for complex scenes that is parameterized by solar, sensor, and scene geometry to support vehicle detection tasks.

• A visible RGB division-of-focal-plane (DoFP) polarimeter is used to collect all data to allow for estimation of the spectro-polarimetric Stokes vector image data.
• A detailed terrain model of a parking lot setting was accurately designed to 1:64 scale that was 3D printed, painted, and based with modeling materials.
• Model vehicles were used to create 28 unique scenarios in dense (4), moderate (8), and sparse (16) configurations.
• All experiments are designed and autonomously conducted in the ASL's Automated Remote Sensing Solar Simulation (ARSSS) laboratory at the University of Dayton.
• The dataset contains 14,112 RGB Stokes vector images (provided in raw RGB DoFP modulated intensity format) across 28 different scenarios, resulting in 4,608 unique scene views, each under 14 different illumination geometries.
• Annotation data is provided for each of the 4,608 different scene views for each model vehicle class.

Source code for using the dataset is available for download from the ASL's code series.

For any use of this dataset, please cite the following:

• DOI for this collection: https://doi.org/10.26890/tji3354
• B. M. Ratliff, "ASL2025: Parameterized Polarimetric Datasets for Remote Sensing Applications," SPIE Optics + Photonics, Polarization Science and Remote Sensing XII, 2025. (Accepted for presentation)

For additional information on the ASL's ARSSS lab at the University of Dayton, please refer to:

• Bradley M. Ratliff "Remote sensing solar simulation laboratory for polarimetric scene simulation", Proc. SPIE 12690, Polarization Science and Remote Sensing XI, 1269002 (3 October 2023); https://doi.org/10.1117/12.2676763.
• B. M. Ratliff, “Development of a laboratory-based scene generation capability to support polarimetric phenomenology studies,” SPIE Henri Poincare Webinar Series on Optical Polarization and Related Phenomena, January 2025. Presentation Link.