Real-Time Object Detection and Tracking Based on Embedded Edge Devices for Local Dynamic Map Generation
This paper proposes a camera system designed for local dynamic map (LDM) generation, capable of simultaneously performing object detection, tracking, and 3D position estimation. This paper focuses on improving existing approaches to better suit our application, rather than proposing novel methods. W...
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2024-02-01
|
Series: | Electronics |
Subjects: | |
Online Access: | https://www.mdpi.com/2079-9292/13/5/811 |
_version_ | 1797264670640308224 |
---|---|
author | Kyoungtaek Choi Jongwon Moon Ho Gi Jung Jae Kyu Suhr |
author_facet | Kyoungtaek Choi Jongwon Moon Ho Gi Jung Jae Kyu Suhr |
author_sort | Kyoungtaek Choi |
collection | DOAJ |
description | This paper proposes a camera system designed for local dynamic map (LDM) generation, capable of simultaneously performing object detection, tracking, and 3D position estimation. This paper focuses on improving existing approaches to better suit our application, rather than proposing novel methods. We modified the detection head of YOLOv4 to enhance the detection performance for small objects and to predict fiducial points for 3D position estimation. The modified detector, compared to YOLOv4, shows an improvement of approximately 5% mAP on the Visdrone2019 dataset and around 3% mAP on our database. We also proposed a tracker based on DeepSORT. Unlike DeepSORT, which applies a feature extraction network for each detected object, the proposed tracker applies a feature extraction network once for the entire image. To increase the resolution of feature maps, the tracker integrates the feature aggregation network (FAN) structure into the DeepSORT network. The difference in multiple objects tracking accuracy (MOTA) between the proposed tracker and DeepSORT is minimal at 0.3%. However, the proposed tracker has a consistent computational load, regardless of the number of detected objects, because it extracts a feature map once for the entire image. This characteristic makes it suitable for embedded edge devices. The proposed methods have been implemented on a system on chip (SoC), Qualcomm QCS605, using network pruning and quantization. This enables the entire process to be executed at 10 Hz on this edge device. |
first_indexed | 2024-04-25T00:32:36Z |
format | Article |
id | doaj.art-cc413272481c4685870fba4bfc2b826e |
institution | Directory Open Access Journal |
issn | 2079-9292 |
language | English |
last_indexed | 2024-04-25T00:32:36Z |
publishDate | 2024-02-01 |
publisher | MDPI AG |
record_format | Article |
series | Electronics |
spelling | doaj.art-cc413272481c4685870fba4bfc2b826e2024-03-12T16:42:14ZengMDPI AGElectronics2079-92922024-02-0113581110.3390/electronics13050811Real-Time Object Detection and Tracking Based on Embedded Edge Devices for Local Dynamic Map GenerationKyoungtaek Choi0Jongwon Moon1Ho Gi Jung2Jae Kyu Suhr3Department of Robotics Engineering, Daegu Catholic University, 13-13, Hayang-ro, Hayang-eup, Gyeongsan-si 38430, Republic of KoreaIIR Seeker R&D Center, LIG Nex1, Mabuk-ro 207, Giheung-gu, Yongin-si 16911, Republic of KoreaDepartment of Electronic Engineering, Korea National University of Transportation, 50 Daehak-ro, Chungju-si 27469, Republic of KoreaDepartment of Intelligent Mechatronics Engineering, Sejong University, 209 Neungdong-ro, Gwangjin-gu, Seoul 05006, Republic of KoreaThis paper proposes a camera system designed for local dynamic map (LDM) generation, capable of simultaneously performing object detection, tracking, and 3D position estimation. This paper focuses on improving existing approaches to better suit our application, rather than proposing novel methods. We modified the detection head of YOLOv4 to enhance the detection performance for small objects and to predict fiducial points for 3D position estimation. The modified detector, compared to YOLOv4, shows an improvement of approximately 5% mAP on the Visdrone2019 dataset and around 3% mAP on our database. We also proposed a tracker based on DeepSORT. Unlike DeepSORT, which applies a feature extraction network for each detected object, the proposed tracker applies a feature extraction network once for the entire image. To increase the resolution of feature maps, the tracker integrates the feature aggregation network (FAN) structure into the DeepSORT network. The difference in multiple objects tracking accuracy (MOTA) between the proposed tracker and DeepSORT is minimal at 0.3%. However, the proposed tracker has a consistent computational load, regardless of the number of detected objects, because it extracts a feature map once for the entire image. This characteristic makes it suitable for embedded edge devices. The proposed methods have been implemented on a system on chip (SoC), Qualcomm QCS605, using network pruning and quantization. This enables the entire process to be executed at 10 Hz on this edge device.https://www.mdpi.com/2079-9292/13/5/811edge computingnetwork slimming embedded devicesobject detectorobject trackerQualcomm SoCcamera for ITS |
spellingShingle | Kyoungtaek Choi Jongwon Moon Ho Gi Jung Jae Kyu Suhr Real-Time Object Detection and Tracking Based on Embedded Edge Devices for Local Dynamic Map Generation Electronics edge computing network slimming embedded devices object detector object tracker Qualcomm SoC camera for ITS |
title | Real-Time Object Detection and Tracking Based on Embedded Edge Devices for Local Dynamic Map Generation |
title_full | Real-Time Object Detection and Tracking Based on Embedded Edge Devices for Local Dynamic Map Generation |
title_fullStr | Real-Time Object Detection and Tracking Based on Embedded Edge Devices for Local Dynamic Map Generation |
title_full_unstemmed | Real-Time Object Detection and Tracking Based on Embedded Edge Devices for Local Dynamic Map Generation |
title_short | Real-Time Object Detection and Tracking Based on Embedded Edge Devices for Local Dynamic Map Generation |
title_sort | real time object detection and tracking based on embedded edge devices for local dynamic map generation |
topic | edge computing network slimming embedded devices object detector object tracker Qualcomm SoC camera for ITS |
url | https://www.mdpi.com/2079-9292/13/5/811 |
work_keys_str_mv | AT kyoungtaekchoi realtimeobjectdetectionandtrackingbasedonembeddededgedevicesforlocaldynamicmapgeneration AT jongwonmoon realtimeobjectdetectionandtrackingbasedonembeddededgedevicesforlocaldynamicmapgeneration AT hogijung realtimeobjectdetectionandtrackingbasedonembeddededgedevicesforlocaldynamicmapgeneration AT jaekyusuhr realtimeobjectdetectionandtrackingbasedonembeddededgedevicesforlocaldynamicmapgeneration |