Feature-Enhanced CenterNet for Small Object Detection in Remote Sensing Images
Compared with anchor-based detectors, anchor-free detectors have the advantage of flexibility and a lower calculation complexity. However, in complex remote sensing scenes, the limited geometric size, weak features of objects, and widely distributed environmental elements similar to the characterist...
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-10-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/14/21/5488 |
| _version_ | 1797466598125076480 |
|---|---|
| author | Tianjun Shi Jinnan Gong Jianming Hu Xiyang Zhi Wei Zhang Yin Zhang Pengfei Zhang Guangzheng Bao |
| author_facet | Tianjun Shi Jinnan Gong Jianming Hu Xiyang Zhi Wei Zhang Yin Zhang Pengfei Zhang Guangzheng Bao |
| author_sort | Tianjun Shi |
| collection | DOAJ |
| description | Compared with anchor-based detectors, anchor-free detectors have the advantage of flexibility and a lower calculation complexity. However, in complex remote sensing scenes, the limited geometric size, weak features of objects, and widely distributed environmental elements similar to the characteristics of objects make small object detection a challenging task. To solve these issues, we propose an anchor-free detector named FE-CenterNet, which can accurately detect small objects such as vehicles in complicated remote sensing scenes. First, we designed a feature enhancement module (FEM) composed of a feature aggregation structure (FAS) and an attention generation structure (AGS). This module contributes to suppressing the interference of false alarms in the scene by mining multiscale contextual information and combining a coordinate attention mechanism, thus improving the perception of small objects. Meanwhile, to meet the high positioning accuracy requirements of small objects, we proposed a new loss function without extra calculation and time cost during the inference process. Finally, to verify the algorithm performance and provide a foundation for subsequent research, we established a dim and small vehicle dataset (DSVD) containing various objects and complex scenes. The experiment results demonstrate that the proposed method performs better than mainstream object detectors. Specifically, the average precision (AP) metric of our method is 7.2% higher than that of the original CenterNet with only a decrease of 1.3 FPS. |
| first_indexed | 2024-03-09T18:42:02Z |
| format | Article |
| id | doaj.art-683e394220884a19af9473f0db1e523d |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-09T18:42:02Z |
| publishDate | 2022-10-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-683e394220884a19af9473f0db1e523d2023-11-24T06:39:51ZengMDPI AGRemote Sensing2072-42922022-10-011421548810.3390/rs14215488Feature-Enhanced CenterNet for Small Object Detection in Remote Sensing ImagesTianjun Shi0Jinnan Gong1Jianming Hu2Xiyang Zhi3Wei Zhang4Yin Zhang5Pengfei Zhang6Guangzheng Bao7Research Center for Space Optical Engineering, Harbin Institute of Technology, Harbin 150001, ChinaResearch Center for Space Optical Engineering, Harbin Institute of Technology, Harbin 150001, ChinaResearch Center for Space Optical Engineering, Harbin Institute of Technology, Harbin 150001, ChinaResearch Center for Space Optical Engineering, Harbin Institute of Technology, Harbin 150001, ChinaResearch Center for Space Optical Engineering, Harbin Institute of Technology, Harbin 150001, ChinaKey Laboratory of Space Photoelectric Detection and Perception (Nanjing University of Aeronautics and Astronautics), Ministry of Industry and Information Technology, Nanjing 211106, ChinaResearch Center for Space Optical Engineering, Harbin Institute of Technology, Harbin 150001, ChinaResearch Center for Space Optical Engineering, Harbin Institute of Technology, Harbin 150001, ChinaCompared with anchor-based detectors, anchor-free detectors have the advantage of flexibility and a lower calculation complexity. However, in complex remote sensing scenes, the limited geometric size, weak features of objects, and widely distributed environmental elements similar to the characteristics of objects make small object detection a challenging task. To solve these issues, we propose an anchor-free detector named FE-CenterNet, which can accurately detect small objects such as vehicles in complicated remote sensing scenes. First, we designed a feature enhancement module (FEM) composed of a feature aggregation structure (FAS) and an attention generation structure (AGS). This module contributes to suppressing the interference of false alarms in the scene by mining multiscale contextual information and combining a coordinate attention mechanism, thus improving the perception of small objects. Meanwhile, to meet the high positioning accuracy requirements of small objects, we proposed a new loss function without extra calculation and time cost during the inference process. Finally, to verify the algorithm performance and provide a foundation for subsequent research, we established a dim and small vehicle dataset (DSVD) containing various objects and complex scenes. The experiment results demonstrate that the proposed method performs better than mainstream object detectors. Specifically, the average precision (AP) metric of our method is 7.2% higher than that of the original CenterNet with only a decrease of 1.3 FPS.https://www.mdpi.com/2072-4292/14/21/5488small object detectionremote sensingCenterNet frameworkmultiscale informationattention mechanism |
| spellingShingle | Tianjun Shi Jinnan Gong Jianming Hu Xiyang Zhi Wei Zhang Yin Zhang Pengfei Zhang Guangzheng Bao Feature-Enhanced CenterNet for Small Object Detection in Remote Sensing Images Remote Sensing small object detection remote sensing CenterNet framework multiscale information attention mechanism |
| title | Feature-Enhanced CenterNet for Small Object Detection in Remote Sensing Images |
| title_full | Feature-Enhanced CenterNet for Small Object Detection in Remote Sensing Images |
| title_fullStr | Feature-Enhanced CenterNet for Small Object Detection in Remote Sensing Images |
| title_full_unstemmed | Feature-Enhanced CenterNet for Small Object Detection in Remote Sensing Images |
| title_short | Feature-Enhanced CenterNet for Small Object Detection in Remote Sensing Images |
| title_sort | feature enhanced centernet for small object detection in remote sensing images |
| topic | small object detection remote sensing CenterNet framework multiscale information attention mechanism |
| url | https://www.mdpi.com/2072-4292/14/21/5488 |
| work_keys_str_mv | AT tianjunshi featureenhancedcenternetforsmallobjectdetectioninremotesensingimages AT jinnangong featureenhancedcenternetforsmallobjectdetectioninremotesensingimages AT jianminghu featureenhancedcenternetforsmallobjectdetectioninremotesensingimages AT xiyangzhi featureenhancedcenternetforsmallobjectdetectioninremotesensingimages AT weizhang featureenhancedcenternetforsmallobjectdetectioninremotesensingimages AT yinzhang featureenhancedcenternetforsmallobjectdetectioninremotesensingimages AT pengfeizhang featureenhancedcenternetforsmallobjectdetectioninremotesensingimages AT guangzhengbao featureenhancedcenternetforsmallobjectdetectioninremotesensingimages |