README.md

metadata

license: mit
task_categories:
  - image-segmentation
  - image-classification
language:
  - en
tags:
  - code
pretty_name: RoadLine Marking Dataset
size_categories:
  - 1K<n<10K

Road Line and Marking Segmentation Dataset (RLMD)

This repository contains dataset and additional information for paper RLMD: A Dataset for Road Marking Segmentation.

Table of Contents

1. Introduction
2. Citation

Introduction

RLMD is a road line and marking semantic segmentation dataset containing 2137 driving scene images and annotations. The annotations is manually annotated with 25 categories and saved in polygon mask format. Information about the categories is shown bellow, or you can download the [csv].

id	name	abbr	r	g	b
0	background	BG	0	0	0
1	box junction	BJ	255	242	0
2	crosswalk	CW	34	117	76
3	stop line	SL	61	72	204
4	solid single white	SSW	237	28	36
5	solid single yellow	SSY	163	73	164
6	solid single red	SSR	185	122	87
7	solid double white	SDW	136	0	21
8	solid double yellow	SDY	112	146	190
9	dashed single white	DSW	181	230	29
10	dashed single yellow	DSY	153	217	234
11	left arrow	LA	158	159	76
12	straight arrow	SA	121	138	134
13	right arrow	RA	41	64	96
14	left straight arrow	LSA	7	102	146
15	right straight arrow	RSA	247	153	255
16	channelizing line	CL	255	204	153
17	motor prohibited	MP	155	255	153
18	slow	SLOW	255	153	173
19	motor priority lane	MPL	230	224	147
20	motor waiting zone	MWZ	35	27	87
21	left turn box	LTB	193	158	155
22	motor icon	MI	109	29	78
23	bike icon	BI	3	164	204
24	parking lot	PL	175	157	185

Citation

This a dataset curated and labelled in proceeding of the following paper, please cite the following paper.

@inproceedings{
hsiao2023rlmd,
title={RLMD: A Dataset for Road Marking Segmentation},
author={Hsiao, Heng-Chih and Cai, Yi-Chang and Lin, Huei-Yung and Chiu, Wei-Chen and Chan, Chiao-Tung},
booktitle={2023 International Conference on Consumer Electronics-Taiwan (ICCE-Taiwan)},
pages={427--428},
year={2023},
organization={IEEE}
}

---

license: mit task_categories: - image-segmentation - image-feature-extraction language: - en tags: - code