CA-YOLO: Model Optimization for Remote Sensing Image Object Detection
The CA-YOLO (Coordinate Attention-YOLO) model has been optimized for object detection in complex remote sensing images, addressing key issues faced by algorithms that detect multiple objects. These issues include weak multi-scale feature learning capabilities and the challenging trade-off between de...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2023-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10167625/ |
| _version_ | 1797787386891993088 |
|---|---|
| author | Lingyun Shen Baihe Lang Zhengxun Song |
| author_facet | Lingyun Shen Baihe Lang Zhengxun Song |
| author_sort | Lingyun Shen |
| collection | DOAJ |
| description | The CA-YOLO (Coordinate Attention-YOLO) model has been optimized for object detection in complex remote sensing images, addressing key issues faced by algorithms that detect multiple objects. These issues include weak multi-scale feature learning capabilities and the challenging trade-off between detection accuracy and model parameter complexity. The CA-YOLO model, built on the framework of YOLOv5, incorporates a lightweight coordinate attention module in the shallow layer to improve detailed feature extraction and suppress redundant information interference. Additionally, a spatial pyramid pooling-fast with a tandem construction module is implemented in the deeper layer. The model employs a stochastic pooling strategy to fuse multi-scale key feature information from low-level to high-level layers, reducing the number of model parameters while improving inference speed. We optimized the anchor box mechanism and modified loss function to improve the ability of the model to detect objects of different sizes and scales. Results show that the CA-YOLO model outperforms the original YOLO in terms of multi-object detection accuracy, with an average mAP@0.5 accuracy improvement of 4.8% and mAP@0.5:0.95 accuracy improvement of 3.8%. Additionally, the CA-YOLO model demonstrates exceptional inference speed, averaging 125 fps, which reinforces its superiority in detection accuracy, generalization ability, and overall efficiency. Notably, these improvements were achieved while maintaining the same number of parameters and complexity as other models, making the CA-YOLO model an exceptional choice for various applications. |
| first_indexed | 2024-03-13T01:21:13Z |
| format | Article |
| id | doaj.art-3f3a48afa792441397c770c0f1686905 |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-03-13T01:21:13Z |
| publishDate | 2023-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-3f3a48afa792441397c770c0f16869052023-07-04T23:00:34ZengIEEEIEEE Access2169-35362023-01-0111647696478110.1109/ACCESS.2023.329048010167625CA-YOLO: Model Optimization for Remote Sensing Image Object DetectionLingyun Shen0https://orcid.org/0009-0001-7107-8169Baihe Lang1Zhengxun Song2Department of Electronic Engineering, Taiyuan Institute of Technology, Taiyuan, ChinaSchool of Electronics and Information Engineering, Changchun University of Science and Technology, Changchun, ChinaOverseas Expertise Introduction Project for Discipline Innovation No. D17017, Changchun University of Science and Technology, Changchun, ChinaThe CA-YOLO (Coordinate Attention-YOLO) model has been optimized for object detection in complex remote sensing images, addressing key issues faced by algorithms that detect multiple objects. These issues include weak multi-scale feature learning capabilities and the challenging trade-off between detection accuracy and model parameter complexity. The CA-YOLO model, built on the framework of YOLOv5, incorporates a lightweight coordinate attention module in the shallow layer to improve detailed feature extraction and suppress redundant information interference. Additionally, a spatial pyramid pooling-fast with a tandem construction module is implemented in the deeper layer. The model employs a stochastic pooling strategy to fuse multi-scale key feature information from low-level to high-level layers, reducing the number of model parameters while improving inference speed. We optimized the anchor box mechanism and modified loss function to improve the ability of the model to detect objects of different sizes and scales. Results show that the CA-YOLO model outperforms the original YOLO in terms of multi-object detection accuracy, with an average mAP@0.5 accuracy improvement of 4.8% and mAP@0.5:0.95 accuracy improvement of 3.8%. Additionally, the CA-YOLO model demonstrates exceptional inference speed, averaging 125 fps, which reinforces its superiority in detection accuracy, generalization ability, and overall efficiency. Notably, these improvements were achieved while maintaining the same number of parameters and complexity as other models, making the CA-YOLO model an exceptional choice for various applications.https://ieeexplore.ieee.org/document/10167625/Object detectionattention mechanismcoordinate attentionSPPFSIoU loss |
| spellingShingle | Lingyun Shen Baihe Lang Zhengxun Song CA-YOLO: Model Optimization for Remote Sensing Image Object Detection IEEE Access Object detection attention mechanism coordinate attention SPPF SIoU loss |
| title | CA-YOLO: Model Optimization for Remote Sensing Image Object Detection |
| title_full | CA-YOLO: Model Optimization for Remote Sensing Image Object Detection |
| title_fullStr | CA-YOLO: Model Optimization for Remote Sensing Image Object Detection |
| title_full_unstemmed | CA-YOLO: Model Optimization for Remote Sensing Image Object Detection |
| title_short | CA-YOLO: Model Optimization for Remote Sensing Image Object Detection |
| title_sort | ca yolo model optimization for remote sensing image object detection |
| topic | Object detection attention mechanism coordinate attention SPPF SIoU loss |
| url | https://ieeexplore.ieee.org/document/10167625/ |
| work_keys_str_mv | AT lingyunshen cayolomodeloptimizationforremotesensingimageobjectdetection AT baihelang cayolomodeloptimizationforremotesensingimageobjectdetection AT zhengxunsong cayolomodeloptimizationforremotesensingimageobjectdetection |