Automated Scene Flow Data Generation for Training and Verification

TL;DR

This paper presents a method to generate synthetic scene flow data with dense ground truth, aiding the training and verification of autonomous driving algorithms where real data is scarce.

Contribution

The authors introduce a novel technology for creating synthetic scene flow datasets with dense and accurate ground truth for autonomous driving applications.

Findings

Generated synthetic scene flow data is dense and precise.

The method supports training and verification of autonomous driving algorithms.

Facilitates development without reliance on real-world data.

Abstract

Scene flow describes the 3D position as well as the 3D motion of each pixel in an image. Such algorithms are the basis for many state-of-the-art autonomous or automated driving functions. For verification and training large amounts of ground truth data is required, which is not available for real data. In this paper, we demonstrate a technology to create synthetic data with dense and precise scene flow ground truth.