HTDet: A Hybrid Transformer-Based Approach for Underwater Small Object Detection
As marine observation technology develops rapidly, underwater optical image object detection is beginning to occupy an important role in many tasks, such as naval coastal defense tasks, aquaculture, etc. However, in the complex marine environment, the images captured by an optical imaging system are...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-02-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/15/4/1076 |
| _version_ | 1797618351930867712 |
|---|---|
| author | Gangqi Chen Zhaoyong Mao Kai Wang Junge Shen |
| author_facet | Gangqi Chen Zhaoyong Mao Kai Wang Junge Shen |
| author_sort | Gangqi Chen |
| collection | DOAJ |
| description | As marine observation technology develops rapidly, underwater optical image object detection is beginning to occupy an important role in many tasks, such as naval coastal defense tasks, aquaculture, etc. However, in the complex marine environment, the images captured by an optical imaging system are usually severely degraded. Therefore, how to detect objects accurately and quickly under such conditions is a critical problem that needs to be solved. In this manuscript, a novel framework for underwater object detection based on a hybrid transformer network is proposed. First, a lightweight hybrid transformer-based network is presented that can extract global contextual information. Second, a fine-grained feature pyramid network is used to overcome the issues of feeble signal disappearance. Third, the test-time-augmentation method is applied for inference without introducing additional parameters. Extensive experiments have shown that the approach we have proposed is able to detect feeble and small objects in an efficient and effective way. Furthermore, our model significantly outperforms the latest advanced detectors with respect to both the number of parameters and the mAP by a considerable margin. Specifically, our detector outperforms the baseline model by 6.3 points, and the model parameters are reduced by 28.5 M. |
| first_indexed | 2024-03-11T08:12:49Z |
| format | Article |
| id | doaj.art-6178615cc5744b8e9699c50e64ec4114 |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-11T08:12:49Z |
| publishDate | 2023-02-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-6178615cc5744b8e9699c50e64ec41142023-11-16T23:03:23ZengMDPI AGRemote Sensing2072-42922023-02-01154107610.3390/rs15041076HTDet: A Hybrid Transformer-Based Approach for Underwater Small Object DetectionGangqi Chen0Zhaoyong Mao1Kai Wang2Junge Shen3School of Marine Science and Technology, Northwestern Polytechnical University, Xi’an 710072, ChinaUnmanned System Research Institute, Northwestern Polytechnical University, Xi’an 710072, ChinaHenan Key Laboratory of Underwater Intelligent Equipment, Zhengzhou 710072, ChinaUnmanned System Research Institute, Northwestern Polytechnical University, Xi’an 710072, ChinaAs marine observation technology develops rapidly, underwater optical image object detection is beginning to occupy an important role in many tasks, such as naval coastal defense tasks, aquaculture, etc. However, in the complex marine environment, the images captured by an optical imaging system are usually severely degraded. Therefore, how to detect objects accurately and quickly under such conditions is a critical problem that needs to be solved. In this manuscript, a novel framework for underwater object detection based on a hybrid transformer network is proposed. First, a lightweight hybrid transformer-based network is presented that can extract global contextual information. Second, a fine-grained feature pyramid network is used to overcome the issues of feeble signal disappearance. Third, the test-time-augmentation method is applied for inference without introducing additional parameters. Extensive experiments have shown that the approach we have proposed is able to detect feeble and small objects in an efficient and effective way. Furthermore, our model significantly outperforms the latest advanced detectors with respect to both the number of parameters and the mAP by a considerable margin. Specifically, our detector outperforms the baseline model by 6.3 points, and the model parameters are reduced by 28.5 M.https://www.mdpi.com/2072-4292/15/4/1076deep learningunderwater object detectiontransformerlightweightfeeble and small object |
| spellingShingle | Gangqi Chen Zhaoyong Mao Kai Wang Junge Shen HTDet: A Hybrid Transformer-Based Approach for Underwater Small Object Detection Remote Sensing deep learning underwater object detection transformer lightweight feeble and small object |
| title | HTDet: A Hybrid Transformer-Based Approach for Underwater Small Object Detection |
| title_full | HTDet: A Hybrid Transformer-Based Approach for Underwater Small Object Detection |
| title_fullStr | HTDet: A Hybrid Transformer-Based Approach for Underwater Small Object Detection |
| title_full_unstemmed | HTDet: A Hybrid Transformer-Based Approach for Underwater Small Object Detection |
| title_short | HTDet: A Hybrid Transformer-Based Approach for Underwater Small Object Detection |
| title_sort | htdet a hybrid transformer based approach for underwater small object detection |
| topic | deep learning underwater object detection transformer lightweight feeble and small object |
| url | https://www.mdpi.com/2072-4292/15/4/1076 |
| work_keys_str_mv | AT gangqichen htdetahybridtransformerbasedapproachforunderwatersmallobjectdetection AT zhaoyongmao htdetahybridtransformerbasedapproachforunderwatersmallobjectdetection AT kaiwang htdetahybridtransformerbasedapproachforunderwatersmallobjectdetection AT jungeshen htdetahybridtransformerbasedapproachforunderwatersmallobjectdetection |