Unsupervised Cluster Guided Object Detection in Aerial Images
Object detection from high-resolution aerial images has received increasing attention during the last few years. It is a common practice to downsize images before feeding them into a network. In real life, there are lots of scenes where many objects gather together in certain areas, such as crossroa...
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2021-01-01
|
| Series: | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9585637/ |
| _version_ | 1798022408533180416 |
|---|---|
| author | Jiajia Liao Yingchao Piao Jinhe Su Guorong Cai Xingwang Huang Long Chen Zhaohong Huang Yundong Wu |
| author_facet | Jiajia Liao Yingchao Piao Jinhe Su Guorong Cai Xingwang Huang Long Chen Zhaohong Huang Yundong Wu |
| author_sort | Jiajia Liao |
| collection | DOAJ |
| description | Object detection from high-resolution aerial images has received increasing attention during the last few years. It is a common practice to downsize images before feeding them into a network. In real life, there are lots of scenes where many objects gather together in certain areas, such as crossroads, parking lots, and playgrounds. The downsizing operation significantly limits the detection ability in these scenes. In this article, we proposed an unsupervised cluster guided detection framework (UCGNet) to address these issues by guiding the detector focus on the object-densely distributed area. In particular, a local location module is first applied to predict a binary map presenting how objects distribute in terms of the pixel of the map. Then, an unsupervised cluster method is used to produce dense regions. Each adjusted dense region is fed into the detector for object detection. Finally, a global merge module generates the final predict results. Experiments were conducted on two popular aerial image datasets including VisDrone2019 and UAVDT. In both datasets, our proposed method outperforms the existing baseline methods with achieving 32.8% and 19.1% mAP, respectively. |
| first_indexed | 2024-04-11T17:29:26Z |
| format | Article |
| id | doaj.art-dbad4eb4dcbb43899a62e37730f387b9 |
| institution | Directory Open Access Journal |
| issn | 2151-1535 |
| language | English |
| last_indexed | 2024-04-11T17:29:26Z |
| publishDate | 2021-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing |
| spelling | doaj.art-dbad4eb4dcbb43899a62e37730f387b92022-12-22T04:12:11ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing2151-15352021-01-0114112041121610.1109/JSTARS.2021.31221529585637Unsupervised Cluster Guided Object Detection in Aerial ImagesJiajia Liao0Yingchao Piao1Jinhe Su2https://orcid.org/0000-0003-1707-5685Guorong Cai3Xingwang Huang4https://orcid.org/0000-0003-3055-6408Long Chen5Zhaohong Huang6Yundong Wu7School of Computer Engineering, Jimei University, Xiamen, ChinaComputer Network Information Center, Chinese Academy of Sciences, Beijing, ChinaSchool of Computer Engineering, Jimei University, Xiamen, ChinaSchool of Computer Engineering, Jimei University, Xiamen, ChinaSchool of Computer Engineering, Jimei University, Xiamen, ChinaSchool of Computer Engineering, Jimei University, Xiamen, ChinaSchool of Computer Engineering, Jimei University, Xiamen, ChinaSchool of Computer Engineering, Jimei University, Xiamen, ChinaObject detection from high-resolution aerial images has received increasing attention during the last few years. It is a common practice to downsize images before feeding them into a network. In real life, there are lots of scenes where many objects gather together in certain areas, such as crossroads, parking lots, and playgrounds. The downsizing operation significantly limits the detection ability in these scenes. In this article, we proposed an unsupervised cluster guided detection framework (UCGNet) to address these issues by guiding the detector focus on the object-densely distributed area. In particular, a local location module is first applied to predict a binary map presenting how objects distribute in terms of the pixel of the map. Then, an unsupervised cluster method is used to produce dense regions. Each adjusted dense region is fed into the detector for object detection. Finally, a global merge module generates the final predict results. Experiments were conducted on two popular aerial image datasets including VisDrone2019 and UAVDT. In both datasets, our proposed method outperforms the existing baseline methods with achieving 32.8% and 19.1% mAP, respectively.https://ieeexplore.ieee.org/document/9585637/Aerial imageconvolutional neural networks (CNNs)deep learningobject detectionunmanned aerial vehicles |
| spellingShingle | Jiajia Liao Yingchao Piao Jinhe Su Guorong Cai Xingwang Huang Long Chen Zhaohong Huang Yundong Wu Unsupervised Cluster Guided Object Detection in Aerial Images IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing Aerial image convolutional neural networks (CNNs) deep learning object detection unmanned aerial vehicles |
| title | Unsupervised Cluster Guided Object Detection in Aerial Images |
| title_full | Unsupervised Cluster Guided Object Detection in Aerial Images |
| title_fullStr | Unsupervised Cluster Guided Object Detection in Aerial Images |
| title_full_unstemmed | Unsupervised Cluster Guided Object Detection in Aerial Images |
| title_short | Unsupervised Cluster Guided Object Detection in Aerial Images |
| title_sort | unsupervised cluster guided object detection in aerial images |
| topic | Aerial image convolutional neural networks (CNNs) deep learning object detection unmanned aerial vehicles |
| url | https://ieeexplore.ieee.org/document/9585637/ |
| work_keys_str_mv | AT jiajialiao unsupervisedclusterguidedobjectdetectioninaerialimages AT yingchaopiao unsupervisedclusterguidedobjectdetectioninaerialimages AT jinhesu unsupervisedclusterguidedobjectdetectioninaerialimages AT guorongcai unsupervisedclusterguidedobjectdetectioninaerialimages AT xingwanghuang unsupervisedclusterguidedobjectdetectioninaerialimages AT longchen unsupervisedclusterguidedobjectdetectioninaerialimages AT zhaohonghuang unsupervisedclusterguidedobjectdetectioninaerialimages AT yundongwu unsupervisedclusterguidedobjectdetectioninaerialimages |