Synth It Like KITTI: Synthetic Data Generation for Object Detection in Driving Scenarios

Richard Marcus, Christian Vogel, Inga Jatzkowski, Niklas Knoop, Marc Stamminger
2/20/2025
cs.CVcs.RO

No official code found. Best implementation: richardmarcus/synth-it-like-kitti (7 stars, MIT, last commit 2025-10-17).

Abstract

An important factor in advancing autonomous driving systems is simulation. Yet, there is rather small progress for transferability between the virtual and real world. We revisit this problem for 3D object detection on LiDAR point clouds and propose a dataset generation pipeline based on the CARLA simulator. Utilizing domain randomization strategies and careful modeling, we are able to train an object detector on the synthetic data and demonstrate strong generalization capabilities to the KITTI dataset. Furthermore, we compare different virtual sensor variants to gather insights, which sensor attributes can be responsible for the prevalent domain gap. Finally, fine-tuning with a small portion of real data almost matches the baseline and with the full training set slightly surpasses it.

View on arXivView PDF

Code Implementations(1)

richardmarcus/synth-it-like-kittiCommunity100%
MIT

Code and Dataset for the Paper: "Synth It Like KITTI: Synthetic Data Generation for Object Detection in Driving Scenarios"

70Shell, PythonJan 9, 20251 years agoMIT
3dcomputer-visionlidarobject-detectionsimulation

Similar Papers

JiSAM: Alleviate Labeling Burden and Corner Case Problems in Autonomous Driving via Minimal Real-World Data

88% match1 repo52026

Domain Adaptation for Different Sensor Configurations in 3D Object Detection

88% match2025

Evaluating the Impact of Synthetic Data on Object Detection Tasks in Autonomous Driving

87% match2025

SynthDrive: Scalable Real2Sim2Real Sensor Simulation Pipeline for High-Fidelity Asset Generation and Driving Data Synthesis

85% match2025

High-Fidelity Digital Twins for Bridging the Sim2Real Gap in LiDAR-Based ITS Perception

85% match2025

Multimodal 3D Object Detection on Unseen Domains

85% match2024

Cite this paper

@article{marcus2025synth,
  title  = {Synth It Like KITTI: Synthetic Data Generation for Object Detection in Driving Scenarios},
  author = {Richard Marcus and Christian Vogel and Inga Jatzkowski and Niklas Knoop and Marc Stamminger},
  year   = {2025},
  eprint = {2502.15076},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2502.15076v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2502.15076.svg)](https://codeofpaper.com/paper/2502.15076)

Discussion