Multi-modal panoramic 3D outdoor datasets for place categorization

Hojung Jung; Yuki Oto; Oscar M. Mozos; Yumi Iwashita; Ryo Kurazume

cs.ROApr 14, 2026

Multi-modal panoramic 3D outdoor datasets for place categorization

Authors: Hojung Jung, Yuki Oto, Oscar M. Mozos, Yumi Iwashita, Ryo Kurazume

Organizations: Graduate School of Information Science and Electrical Engineering, Kyushu University, Fukuoka 819-0395, Japan · Polytechnic University of Cartagena (UPCT), Spain · Faculty of Information Science and Electrical Engineering, Kyushu University, Fukuoka 819-0395, Japan

Abstract

We present two multi-modal panoramic 3D outdoor (MPO) datasets for semantic place categorization with six categories: forest, coast, residential area, urban area and indoor/outdoor parking lot. The first dataset consists of 650 static panoramic scans of dense (9,000,000 points) 3D color and reflectance point clouds obtained using a FARO laser scanner with synchronized color images. The second dataset consists of 34,200 real-time panoramic scans of sparse (70,000 points) 3D reflectance point clouds obtained using a Velodyne laser scanner while driving a car. The datasets were obtained in the city of Fukuoka, Japan and are publicly available in [1], [2]. In addition, we compare several approaches for semantic place categorization with best results of 96.42% (dense) and 89.67% (sparse).

Figures & tables

Fig. 1: An example map of MPO Dataset with six place categories: (1) forest, (2) coast, (3) indoor parking lot, (4) outdoor parking lot, (5) residential area and (6) urban area

Fig. 2: Experimental setup for Dense MPO Dataset equipped with (1) a FARO Focus3D sensor system and for Sparse MPO Dataset equipped with (2) a Velodyne HDL-32E laser scanner, (3) a Kodak PIXPRO SP360 camera and (4) a GARMIN GPS 18x LVC

Category	Number of scans by location							Total
Category	Set1	Set2	Set3	Set4	Set5	Set6	Set7	Total
Coast	14	14	16	12	17	14	16	103
Forest	16	16	17	18	16	16	17	116
Parking lot (in)	16	16	13	15	17	13	13	105
Parking lot (out)	15	17	16	15	15	14	16	108
Residential area	14	16	14	15	16	15	16	106

TABLE I: DENSE MPO DATASET of outdoor scene containing 650 pairs of range, reflectance and color images

Panel of Figure 4 — Fig. 3: Dense MPO Dataset : examples of high-resolution range, reflectance and color panoramic images for six outdoor place categories: forest, coast, indoor/outdoor parking lot, residential and urban area. In range images, darker colors indicate closer distances and in reflectance images, brighter colors indicate higher intensity.

Category	Number of scans by location					Total
	Set1	Set2	Set3	Set4	Set5
	Set6	Set7	Set8	Set9	Set10
Coast	511	254	571	221	314	4298
Coast	376	872	506	386	287	4298
Forest	440	824	980	707	730	6479
Forest	720	439	311	797	531	6479

TABLE II: SPARSE MPO DATASET of outdoor scene containing 34,200 pairs of range and reflectance scans, GPS position data and color images

Panel of Figure 6 — Fig. 4: Sparse MPO Dataset : examples of low-resolution range and reflectance panoramic images for six outdoor place categories: ‘forest’, ‘coast’, ‘indoor parking lot’, ‘outdoor parking lot’, ‘residential area’ and ‘urban area’. In range images, darker colors indicate closer distances and in reflectance images, brighter colors indicate higher intensity.

Descriptor	Range	Reflectance	Color
Spin image [ 14 ]	$89.43\pm 0.00$	-	-
LBP [ 13 ]	$94.35\pm 2.67$	$\mathbf{96.42\pm 2.68}$	$93.86\pm 3.85$
Texton [ 15 ]	$89.04\pm 5.58$	$73.59\pm 13.73$	$81.81\pm 10.67$

TABLE III: Correct classification ratio (CCR) results of standard descriptors with single modality using DENSE MPO DATASET [%]

	Coast	Forest	ParkingIn	ParkingOut	Residential	Urban
Coast	$\mathbf{93.45\pm 17.39}$	${6.54\pm 17.39}$	${0.00\pm 0.00}$	${0.00\pm 0.00}$	${0.00\pm 0.00}$	${0.00\pm 0.00}$
Forest	${8.95\pm 8.27}$	$\mathbf{91.04\pm 8.27}$	${0.00\pm 0.00}$	${0.00\pm 0.00}$	${0.00\pm 0.00}$	${0.00\pm 0.00}$
ParkingIn	${0.00\pm 0.00}$	${0.00\pm 0.00}$	$\mathbf{100.00\pm 0.00}$	${0.00\pm 0.00}$	${0.00\pm 0.00}$	${0.00\pm 0.00}$
ParkingOut	${1.91\pm 2.92}$	${0.00\pm 0.00}$	${0.00\pm 0.00}$	$\mathbf{98.08\pm 2.92}$	${0.00\pm 0.00}$	${0.00\pm 0.00}$
Residential	${1.33\pm 2.68}$	${0.00\pm 0.00}$	${0.00\pm 0.00}$	${0.00\pm 0.00}$	$\mathbf{97.32\pm 4.43}$	${1.33\pm 4.00}$
Urban	${0.00\pm 0.00}$	${0.00\pm 0.00}$	${0.00\pm 0.00}$	${0.00\pm 0.00}$	${1.87\pm 4.00}$	$\mathbf{98.12\pm 4.00}$

TABLE IV: Confusion matrix of reflectance LBP images for DENSE MPO DATASET [CCR%]

Modality	Range + Reflectance
Descriptor	LBP [ 10 ]	LTP [ 12 ]
CCR[%]	$95.67\pm 3.69$	$92.84\pm 3.33$

TABLE V: Correct classification ratio (CCR) results of multi-modalities using DENSE MPO DATASET

Descriptor	Technique
Descriptor	None	Majority vote
Spin image [ 14 ]	$79.23\pm 4.51$	$88.34\pm 0.12$
LBP [ 13 ]	$83.98\pm 4.59$	$\mathbf{89.67\pm 0.21}$

TABLE VI: Correct classification ratio (CCR) results of standard descriptors and majority vote using SPARSE MPO DATASET [%]

Fig. 5: A performance of Sparse MPO Dataset by applying a majority vote technique

	Coast	Forest	ParkingIn	ParkingOut	Residential	Urban
Coast	$\mathbf{82.99\pm 4.19}$	${5.20\pm 0.54}$	${0.00\pm 0.00}$	${7.15\pm 2.20}$	${1.50\pm 0.05}$	${3.16\pm 0.56}$
Forest	${8.64\pm 1.12}$	$\mathbf{90.87\pm 1.20}$	${0.00\pm 0.00}$	${0.49\pm 0.01}$	${0.00\pm 0.00}$	${0.00\pm 0.00}$
ParkingIn	${0.00\pm 0.00}$	${3.31\pm 0.01}$	$\mathbf{93.95\pm 0.65}$	${2.16\pm 0.56}$	${0.59\pm 0.02}$	${0.00\pm 0.00}$
ParkingOut	${0.96\pm 0.06}$	${0.00\pm 0.00}$	${5.25\pm 0.12}$	$\mathbf{89.16\pm 1.31}$	${2.44\pm 0.26}$	${2.19\pm 0.26}$
Residential	${0.45\pm 0.01}$	${0.00\pm 0.00}$	${0.00\pm 0.00}$	${2.70\pm 0.01}$	$\mathbf{93.23\pm 0.20}$	${3.63\pm 0.20}$
Urban	${0.00\pm 0.00}$	${0.00\pm 0.00}$	${0.00\pm 0.00}$	${3.84\pm 0.76}$	${8.76\pm 0.79}$	$\mathbf{87.39\pm 0.20}$

TABLE VII: Confusion matrix of LBP images with majority vote for SPARSE MPO DATASET [CCR%]

3D Scene Understanding

3D scene understanding builds geometric and semantic representations of environments for spatial reasoning, perception, and interaction.

See all 293 papers

Learning Geometric and Photometric Features from Panoramic LiDAR Scans for Outdoor Place Categorization
Mar 13, 2026Kazuto Nakashima, Hojung Jung, Yuki Oto +3Visual Place Recognition 3D Scene Understanding
Graduate School of Information Science and Electrical Engineering, Kyushu University, Fukuoka, Japan · Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, USA · Faculty of Information Science and Electrical Engineering, Kyushu University, Fukuoka, Japan +1
The P $^3$ Dataset: Pixels, Points and Polygons for Multimodal Building Vectorization
May 21, 2025Raphael Sulzer, Liuyun Duan, Nicolas Girard +1Remote Sensing Image Segmentation Building Footprint Extraction
Université Côte d’Azur, INRIA, Sophia-Antipolis, France · LuxCarta Technology, Mouans-Sartoux, France
3D Segmentation Using Viewpoint-Dependent Spatial Relationships
May 15, 2026Ayaka Nanri, Klara Reichard, Mert Kiray +33D Spatial Reasoning 3D Referring Segmentation
Institute of Science Tokyo · Technical University of Munich · BMW Group +12

Mar 13, 2026J/K move · Enter open · S save

Learning Geometric and Photometric Features from Panoramic LiDAR Scans for Outdoor Place Categorization

Kazuto Nakashima, Hojung Jung, Yuki Oto, Yumi Iwashita+2

Graduate School of Information Science and Electrical Engineering, Kyushu University, Fukuoka, Japan · Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, USA · Faculty of Information Science and Electrical Engineering, Kyushu University, Fukuoka, Japan+1

Visual Place Recognition 3D Scene Understanding Point Cloud Classification

Open paper AskPDF arXiv

Semantic place categorization, which is one of the essential tasks for autonomous robots and vehicles, allows them to have capabilities of self-decision and navigation in unfamiliar environments. In particular, outdoor places are more difficult targets than indoor ones due to perceptual variations, such as dynamic illuminance over twenty-four hours and occlusions by cars and pedestrians. This paper presents a novel method of categorizing outdoor places using convolutional neural networks (CNNs), which take omnidirectional depth/reflectance images obtained by 3D LiDARs as the inputs. First, we construct a large-scale outdoor place dataset named Multi-modal Panoramic 3D Outdoor (MPO) comprising two types of point clouds captured by two different LiDARs. They are labeled with six outdoor place categories: coast, forest, indoor/outdoor parking, residential area, and urban area. Second, we provide CNNs for LiDAR-based outdoor place categorization and evaluate our approach with the MPO dataset. Our results on the MPO dataset outperform traditional approaches and show the effectiveness in which we use both depth and reflectance modalities. To analyze our trained deep networks we visualize the learned features.

Abstract

3D Scene Understanding

Explore similar work

Learning Geometric and Photometric Features from Panoramic LiDAR Scans for Outdoor Place Categorization

The P3^33 Dataset: Pixels, Points and Polygons for Multimodal Building Vectorization

3D Segmentation Using Viewpoint-Dependent Spatial Relationships

The P $^3$ Dataset: Pixels, Points and Polygons for Multimodal Building Vectorization