Oscar Martinez Mozos

Fukuoka Datasets for Place Categorization

http://robotics.ait.kyushu-u.ac.jp/kyushu_datasets/

Several multimodal 3D datasets for the categorization of places. In addition to 3D depth information, other modalities like RGB or reflectance images are included. The datasets include indoor and outdoor scenarios obtained in different locations in Fukuoka city, Japan. Outdoor place categories include forest, urban area, indoor parking, outdoor parking, coast areas, and residential area. Indoor place categories include corridor, office, lab, study room, kitchen, and laboratory. The set of depth sensors include SICK laser range finders and RGB-D cameras on a mobile platform, and Velodyne and FARO LIDARS on a car.

Also reachable through the CVonline repository.

Citation:

Mozos, O. M.; Nakashima, K.; Jung, H.; Iwashita, Y.; and Kurazume, R. Fukuoka Datasets for Place Categorization. International Journal of Robotics Research, 38(5): 507-517. March 2019.

Semantic Categorization of Places

Semantic labeling is the problem of assigning labels with semantic meaning to the different places that composed indoor environments. Typical labels include "corridor", "room", "office", "lab", or "doorway". The ability to learn such semantic categories from sensor data enables a mobile robot to extend the representation of the environment facilitating the interaction with humans. We study different approach to categorize places using as input different sensors such as laser scans, vision cameras, RGB-D cameras.

Datasets:

Data sets corresponding to the semantic classifications of places in different indoor environments using 2D laser scans :

https://doi.org/10.5281/zenodo.10600586

Data sets corresponding to the semantic classifications of places in different indoor environments using RGB-D sensors (Kinect camera):

http://robotics.ait.kyushu-u.ac.jp/~kurazume/r-cv-e.html#c10

place_classification_mobile_robot_3_classes.avi

fr079-6classes-hmm2.anim.avi

CoSM_step1.wmv

Object Detection and Categorization in 3D Point Clouds

We work on several approaches to detect and categorize different furniture objects like chairs, tables or sideboards in 3D point clouds in the presence of clutter and occlusion. The detection and categorization of these objects is an important capability for service robots acting in indoor environments, as they play important roles in the tasks to be performed by these robots.

ROS package for furniture classification using the Kinect camera:

http://www.ros.org/wiki/furniture_classification

kitchen_rotational_scan_1024x768_small.avi

People Detection in Range Data

The recognition of people is key a capacity for service robots that interact with humans. We work on the detection of people with laser scans using one or several layers. The key idea is to learn classifiers for body parts that are represented by segments in laser scans. The main goal is to obtain a robust classifier which is able to detect people in different environments.

ROS package for people detection using range data. Available at the public tum-ros repository under perception/people_detector_node/:

http://tum-ros-pkg.svn.sourceforge.net/

AdaBoostOffice.avi

multi_layer_people_detection_test_1.avi

V-Proportion for cell patterns

The visual system plays a predominant role in the human perception. Although all components of the eye are important to perceive visual information, the retina is a fundamental part of the visual system. In this work we study the spatial relations between neuronal mosaics in the retina. These relations have shown its importance to investigate possible constraints or connectivities between different spatially colocalized populations of neurons, and to explain how visual information spreads along the layers before being sent to the brain. We introduce the V-Proportion, a method based on the Voronoi diagram to study possible spatial interactions between two neuronal mosaics. Results in simulations as well as in real data demonstrate the effectiveness of this method to detect spatial relations between neurons in different layers.

Paper:

Oscar Martinez Mozos, Jose A. Bolea, Jose M. Ferrandez, Peter K. Ahnelt, Eduardo Fernandez, V-Proportion: A method based on the Voronoi diagram to study spatial relations in neuronal mosaics of the retina, Neurocomputing, Volume 74, Issues 1–3, 2010, Pages 418-427, ISSN 0925-2312, https://doi.org/10.1016/j.neucom.2010.07.020.

Datasets:

https://doi.org/10.5281/zenodo.14052297