Multiple Object Detection Based on Clustering and Deep Learning Methods.

Huu Thu Nguyen,Sejin Lee,Chul Hee Bae,Eon-Ho Lee

doi:10.3390/s20164424

Huu Thu Nguyen, Sejin Lee + Show 2 more

Open Access

https://doi.org/10.3390/s20164424

Copy DOI

Journal: Sensors (Basel, Switzerland)	Publication Date: Aug 7, 2020
Citations: 26	License type: CC BY 4.0

Affiliation: Kongju National University

Abstract

Multiple object detection is challenging yet crucial in computer vision. In This study, owing to the negative effect of noise on multiple object detection, two clustering algorithms are used on both underwater sonar images and three-dimensional point cloud LiDAR data to study and improve the performance result. The outputs from using deep learning methods on both types of data are treated with K-Means clustering and density-based spatial clustering of applications with noise (DBSCAN) algorithms to remove outliers, detect and cluster meaningful data, and improve the result of multiple object detections. Results indicate the potential application of the proposed method in the fields of object detection, autonomous driving system, and so forth.

Highlights

The detection of multiple objects is crucial in image processing
The fully convolutional network (FCN) [1] is used for underwater sonar data to detect the presence of the human body and other objects in the underwater environment and to segment those objects
The 3D point cloud LiDAR data of the urban environment are first subjected to a spherical signature descriptor (SSD) [2] to change spatial data to color image data; subsequently, the convolutional neural network (CNN) [3] is used on the output data to classify and segment objects in the environment

Summary

Introduction

The detection of multiple objects is crucial in image processing. It has been investigated extensively owing to its potential wide application in numerous fields, such as computer vision, machine inspection, manufacturing industry, and self-driving cars. The 3D point cloud LiDAR data of the urban environment are first subjected to a spherical signature descriptor (SSD) [2] to change spatial data to color image data; subsequently, the convolutional neural network (CNN) [3] is used on the output data to classify and segment objects in the environment. The outputs of both sonar and LiDAR data have a high amount of noise, which result in the low performance results of detection and segmentation tasks. The two scenarios above were managed using two clustering algorithms, that is, K-Means clustering [4] and DBSCAN [5]

Objectives

Methods

Findings

Conclusion