YOLOAL: Focusing on the Object Location for Detection on Drone Imagery
Object detection in drone-captured scenarios, which can be considered as a task of detecting dense small objects, is still a challenge. Drones navigate at different altitudes, causing significant changes in the size of the detected objects and posing a challenge to the model. Additionally, it is nec...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
IEEE
2023-01-01
|
Series: | IEEE Access |
Subjects: | |
Online Access: | https://ieeexplore.ieee.org/document/10318136/ |
_version_ | 1797454687259066368 |
---|---|
author | Xinting Chen Wenzhu Yang Shuang Zeng Lei Geng Yanyan Jiao |
author_facet | Xinting Chen Wenzhu Yang Shuang Zeng Lei Geng Yanyan Jiao |
author_sort | Xinting Chen |
collection | DOAJ |
description | Object detection in drone-captured scenarios, which can be considered as a task of detecting dense small objects, is still a challenge. Drones navigate at different altitudes, causing significant changes in the size of the detected objects and posing a challenge to the model. Additionally, it is necessary to improve the ability of the object detection model to rapidly detect small dense objects. To address these issues, we propose YOLOAL, a model that emphasizes the location information of the objects. It incorporates a new attention mechanism called the Convolution and Coordinate Attention Module (CCAM) into its design. This mechanism performs better than traditional ones in dense small object scenes because it adds coordinates that help identify attention regions in such scenarios. Furthermore, our model uses a new loss function combined with the Efficient IoU (EIoU) and Alpha-IoU methods that achieve better results than the traditional approaches. The proposed model achieved state-of-the-art performance on the VisDrone and DOTA datasets. YOLOAL reaches an AP50 (average accuracy when Intersection over Union threshold is 0.5) of 63.6% and an mAP (average of 10 IoU thresholds, ranging from 0.5 to 0.95) of 40.8% at a real-time speed of 0.27 seconds on the VisDrone dataset, and the mAP on the DOTA dataset even reaches 39% on an NVIDIA A4000. |
first_indexed | 2024-03-09T15:41:42Z |
format | Article |
id | doaj.art-44b1de9ba142428780b00256600f599e |
institution | Directory Open Access Journal |
issn | 2169-3536 |
language | English |
last_indexed | 2024-03-09T15:41:42Z |
publishDate | 2023-01-01 |
publisher | IEEE |
record_format | Article |
series | IEEE Access |
spelling | doaj.art-44b1de9ba142428780b00256600f599e2023-11-25T00:00:57ZengIEEEIEEE Access2169-35362023-01-011112888612889710.1109/ACCESS.2023.333281510318136YOLOAL: Focusing on the Object Location for Detection on Drone ImageryXinting Chen0https://orcid.org/0009-0007-3046-6249Wenzhu Yang1Shuang Zeng2Lei Geng3Yanyan Jiao4School of Cyber Security and Computer, Hebei University, Baoding, ChinaSchool of Cyber Security and Computer, Hebei University, Baoding, ChinaSchool of Cyber Security and Computer, Hebei University, Baoding, ChinaSchool of Cyber Security and Computer, Hebei University, Baoding, ChinaSchool of Cyber Security and Computer, Hebei University, Baoding, ChinaObject detection in drone-captured scenarios, which can be considered as a task of detecting dense small objects, is still a challenge. Drones navigate at different altitudes, causing significant changes in the size of the detected objects and posing a challenge to the model. Additionally, it is necessary to improve the ability of the object detection model to rapidly detect small dense objects. To address these issues, we propose YOLOAL, a model that emphasizes the location information of the objects. It incorporates a new attention mechanism called the Convolution and Coordinate Attention Module (CCAM) into its design. This mechanism performs better than traditional ones in dense small object scenes because it adds coordinates that help identify attention regions in such scenarios. Furthermore, our model uses a new loss function combined with the Efficient IoU (EIoU) and Alpha-IoU methods that achieve better results than the traditional approaches. The proposed model achieved state-of-the-art performance on the VisDrone and DOTA datasets. YOLOAL reaches an AP50 (average accuracy when Intersection over Union threshold is 0.5) of 63.6% and an mAP (average of 10 IoU thresholds, ranging from 0.5 to 0.95) of 40.8% at a real-time speed of 0.27 seconds on the VisDrone dataset, and the mAP on the DOTA dataset even reaches 39% on an NVIDIA A4000.https://ieeexplore.ieee.org/document/10318136/Dronesmall dense objects detectionattention mechanismloss function |
spellingShingle | Xinting Chen Wenzhu Yang Shuang Zeng Lei Geng Yanyan Jiao YOLOAL: Focusing on the Object Location for Detection on Drone Imagery IEEE Access Drone small dense objects detection attention mechanism loss function |
title | YOLOAL: Focusing on the Object Location for Detection on Drone Imagery |
title_full | YOLOAL: Focusing on the Object Location for Detection on Drone Imagery |
title_fullStr | YOLOAL: Focusing on the Object Location for Detection on Drone Imagery |
title_full_unstemmed | YOLOAL: Focusing on the Object Location for Detection on Drone Imagery |
title_short | YOLOAL: Focusing on the Object Location for Detection on Drone Imagery |
title_sort | yoloal focusing on the object location for detection on drone imagery |
topic | Drone small dense objects detection attention mechanism loss function |
url | https://ieeexplore.ieee.org/document/10318136/ |
work_keys_str_mv | AT xintingchen yoloalfocusingontheobjectlocationfordetectionondroneimagery AT wenzhuyang yoloalfocusingontheobjectlocationfordetectionondroneimagery AT shuangzeng yoloalfocusingontheobjectlocationfordetectionondroneimagery AT leigeng yoloalfocusingontheobjectlocationfordetectionondroneimagery AT yanyanjiao yoloalfocusingontheobjectlocationfordetectionondroneimagery |