Saved in:
Bibliographic Details
Main Authors: Mo, Zhaobin, Li, Yunlong, Di, Xuan
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2501.02143
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929658041204736
author Mo, Zhaobin
Li, Yunlong
Di, Xuan
author_facet Mo, Zhaobin
Li, Yunlong
Di, Xuan
contents Safety-critical driving data is crucial for developing safe and trustworthy self-driving algorithms. Due to the scarcity of safety-critical data in naturalistic datasets, current approaches primarily utilize simulated or artificially generated images. However, there remains a gap in authenticity between these generated images and naturalistic ones. We propose a novel framework to augment the safety-critical driving data from the naturalistic dataset to address this issue. In this framework, we first detect vehicles using YOLOv5, followed by depth estimation and 3D transformation to simulate vehicle proximity and critical driving scenarios better. This allows for targeted modification of vehicle dynamics data to reflect potentially hazardous situations. Compared to the simulated or artificially generated data, our augmentation methods can generate safety-critical driving data with minimal compromise on image authenticity. Experiments using KITTI datasets demonstrate that a downstream self-driving algorithm trained on this augmented dataset performs superiorly compared to the baselines, which include SMOGN and importance sampling.
format Preprint
id arxiv_https___arxiv_org_abs_2501_02143
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle SafeAug: Safety-Critical Driving Data Augmentation from Naturalistic Datasets
Mo, Zhaobin
Li, Yunlong
Di, Xuan
Computer Vision and Pattern Recognition
Machine Learning
Safety-critical driving data is crucial for developing safe and trustworthy self-driving algorithms. Due to the scarcity of safety-critical data in naturalistic datasets, current approaches primarily utilize simulated or artificially generated images. However, there remains a gap in authenticity between these generated images and naturalistic ones. We propose a novel framework to augment the safety-critical driving data from the naturalistic dataset to address this issue. In this framework, we first detect vehicles using YOLOv5, followed by depth estimation and 3D transformation to simulate vehicle proximity and critical driving scenarios better. This allows for targeted modification of vehicle dynamics data to reflect potentially hazardous situations. Compared to the simulated or artificially generated data, our augmentation methods can generate safety-critical driving data with minimal compromise on image authenticity. Experiments using KITTI datasets demonstrate that a downstream self-driving algorithm trained on this augmented dataset performs superiorly compared to the baselines, which include SMOGN and importance sampling.
title SafeAug: Safety-Critical Driving Data Augmentation from Naturalistic Datasets
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2501.02143