dataset · Zenodo (CERN European Organization for Nuclear Research)
A road segmentation dataset has been created using dashcam footage to evaluate zero-shot foundation computer vision models. The data was gathered from twenty driving videos that span a broad selection of environments, including rural, mountain, urban, highway, and nighttime settings. Across these recordings, 5,939 frames were extracted at 960 by 720 resolution. The dataset supplies 5,939 segmentation masks generated by the Segment Anything Model ViT-H, accompanied by 151 manually annotated ground truth labels designed for formal evaluation. Its primary objective is to benchmark foundation model performance on unstructured road conditions that are largely absent from conventional urban driving datasets such as Cityscapes and CamVid.
Autonomous navigation and driver-assistance systems must interpret road boundaries accurately outside of well-marked city streets. Conventional benchmarks often overlook challenging rural, mountain, or nighttime routes. Providing dedicated evaluation data for unstructured driving environments helps researchers assess how reliably modern foundation models transfer to diverse, real-world driving conditions without requiring task-specific retraining.
This dataset provides testing infrastructure that could be used by developers of advanced driver-assistance systems and autonomous vehicle software to validate perception algorithms against non-standard road environments. It represents an early-stage benchmarking resource rather than an applied or market-ready commercial product.
AI-generated from the published abstract. Always read the original work before citing.
A dashcam road segmentation dataset collected from 20 driving videos covering diverse road types including rural, mountain, urban, highway, and night conditions. Contains 5,939 extracted frames at 960×720 resolution, 5,939 SAM ViT-H generated road segmentation masks, and 151 manually annotated ground truth labels used for evaluation. Collected to benchmark zero-shot foundation model segmentation on unstructured road environments not represented in standard urban driving datasets such as Cityscapes or CamVid.
This page summarises published work. The authoritative version sits with the publisher.
DOI: 10.5281/zenodo.22301899
Is something wrong with this record? Report it or request removal.
Discussion
Have you built on this work, tried to replicate it, or seen it applied in practice? Share what you know. Verified researchers and MARATTO™ domain experts can open a discussion, and any member can reply. Contributions are reviewed before they appear.
No discussion yet. Open the first thread.
New to MARATTO™? Create a free account.
