WADS 3D Object Detection Dataset (WADS-3D) - sequence 24

Document Type

Data

Publication Date

2-5-2025

Abstract

WADS-3D is a standard KITTI-format dataset designed for analyzing 3D object detection robustness in severe winter weather. Addressing the need for granular analysis of perception failure in snow, the dataset includes 4,100 frames and over 30,000 labeled car instances captured during heavy snowfall and winter-storm conditions. Derived from the large-scale multimodal WADS dataset, WADS-3D enables researchers to move beyond qualitative weather assessments and systematically evaluate how varying heavy snowfall intensities impact a wide range of neural network detectors.

This document is currently not available here.

Share

COinS