Repositorio UVigo

Transfer learning for indoor object classification: from images to point clouds

Investigo Repository

Transfer learning for indoor object classification: from images to point clouds

Balado Frías, Jesús; Díaz Vilariño, Lucía; Verbree, Edward; Arias Sánchez, Pedro
 
DATE : 2020
UNIVERSAL IDENTIFIER : http://hdl.handle.net/11093/1538
UNESCO SUBJECT : 3311.02 Ingeniería de Control ; 3305.22 Metrología de la Edificación
DOCUMENT TYPE : article

ABSTRACT :

Indoor furniture is of great relevance to building occupants in everyday life. Furniture occupies space in the building, gives comfort, establishes order in rooms and locates services and activities. Furniture is not always static; the rooms can be reorganized according to the needs. Keeping the building models up to date with the current furniture is key to work with indoor environments. Laser scanning technology can acquire indoor environments in a fast and precise way, and recent artificial intelligence techniques can classify correctly the objects that contain. The objective of this work is to study how to minimize the use of point cloud samples in Neural Network training, tedious to label, and replace them with images obtained from online sources. For this, point clouds are converted to images by means of rotations and projections. The conversion of a 3D vector data to a 2D raster allows the use of Convolutional Neural Networks, the achievement of several images for each acquired point cloud object and the combination with images obtained from online sources, such as Google ... [+]
Indoor furniture is of great relevance to building occupants in everyday life. Furniture occupies space in the building, gives comfort, establishes order in rooms and locates services and activities. Furniture is not always static; the rooms can be reorganized according to the needs. Keeping the building models up to date with the current furniture is key to work with indoor environments. Laser scanning technology can acquire indoor environments in a fast and precise way, and recent artificial intelligence techniques can classify correctly the objects that contain. The objective of this work is to study how to minimize the use of point cloud samples in Neural Network training, tedious to label, and replace them with images obtained from online sources. For this, point clouds are converted to images by means of rotations and projections. The conversion of a 3D vector data to a 2D raster allows the use of Convolutional Neural Networks, the achievement of several images for each acquired point cloud object and the combination with images obtained from online sources, such as Google Images. The images have been distributed among the validation and testing training sets following different percentages. The results show that, although point cloud images cannot be completely dispensed within the training set, only 10% of these achieve high accuracy in the classification [-]

Show full item record



Files in this item

/ © Author(s) 2020. This work is distributed under the Creative Commons Attribution 4.0 License Except where otherwise noted, this item's license is described as / © Author(s) 2020. This work is distributed under the Creative Commons Attribution 4.0 License
2013 Universidade de Vigo, Todos los derechos reservados
Calidad So9001