Multi-Channel Convolutional Neural Network Based 3D Object Detection for Indoor Robot Environmental Perception

Autor: Hock Soon Seah, Ruifeng Li, Chee Kwang Quah, Jingwen Sun, Budianto Tandianus, Hezi Shi, Lijun Zhao, Li Wang
Přispěvatelé: School of Computer Science and Engineering, School of Electrical and Electronic Engineering, ST Engineering-NTU Corporate Laboratory
Jazyk: angličtina
Rok vydání: 2019
Předmět:
0209 industrial biotechnology
Computer science
ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION
Multi-Channel CNN
02 engineering and technology
lcsh:Chemical technology
Biochemistry
Convolutional neural network
Article
Analytical Chemistry
020901 industrial engineering & automation
0202 electrical engineering
electronic engineering
information engineering
Computer vision
lcsh:TP1-1185
multi-channel CNN
Electrical and Electronic Engineering
Instrumentation
Service robot
Engineering::Computer science and engineering [DRNTU]
environmental perception
business.industry
3D reconstruction
indoor robot
Object (computer science)
Atomic and Molecular Physics
and Optics
Object detection
3D object detection
Feature (computer vision)
RGB color model
Robot
3D Object Detection
020201 artificial intelligence & image processing
Artificial intelligence
Focus (optics)
business
Zdroj: Sensors, Vol 19, Iss 4, p 893 (2019)
Sensors (Basel, Switzerland)
Sensors
Volume 19
Issue 4
ISSN: 1424-8220
Popis: Environmental perception is a vital feature for service robots when working in an indoor environment for a long time. The general 3D reconstruction is a low-level geometric information description that cannot convey semantics. In contrast, higher level perception similar to humans requires more abstract concepts, such as objects and scenes. Moreover, the 2D object detection based on images always fails to provide the actual position and size of an object, which is quite important for a robot&rsquo
s operation. In this paper, we focus on the 3D object detection to regress the object&rsquo
s category, 3D size, and spatial position through a convolutional neural network (CNN). We propose a multi-channel CNN for 3D object detection, which fuses three input channels including RGB, depth, and bird&rsquo
s eye view (BEV) images. We also propose a method to generate 3D proposals based on 2D ones in the RGB image and semantic prior. Training and test are conducted on the modified NYU V2 dataset and SUN RGB-D dataset in order to verify the effectiveness of the algorithm. We also carry out the actual experiments in a service robot to utilize the proposed 3D object detection method to enhance the environmental perception of the robot.
Databáze: OpenAIRE
Externí odkaz:
Nepřihlášeným uživatelům se plný text nezobrazuje