Object detection for drones on Raspberry Pi potentials and challenges

The paper presents preliminary research results about implementing an object detection program on a Single Board Computer. These results are used later to develop applications for drones. The object identification program is developed in Python using the TensorFlow library. The authors have succeede...

Full description

Saved in:
Bibliographic Details
Published in:IOP conference series. Materials Science and Engineering 2021-03, Vol.1109 (1), p.12033
Main Authors: Khoi, Tran Quang, Quang, Nguyen Anh, Hieu, Ngo Khanh
Format: Article
Language:English
Subjects:
Artificial neural networks
Modules
Object recognition
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The paper presents preliminary research results about implementing an object detection program on a Single Board Computer. These results are used later to develop applications for drones. The object identification program is developed in Python using the TensorFlow library. The authors have succeeded in implementing and testing this object identification module using the artificial neural network model SSDMobileNet V2 on the Raspberry Pi 3B+. The results in this paper demonstrate the potential of this module for further development in the future. Based on the simulation and real-world results, the authors showed that a good outcome is achievable with limited resources for the AI module. Along with a high-precision object detection feature, this module can also estimate the distance and velocity of the “human” object with good accuracy. Besides, the paper also proposes several solutions to increase the performance and most importantly, the real-time feature of the developed module.
ISSN:1757-8981
1757-899X
DOI:10.1088/1757-899X/1109/1/012033