Daytime Sea Fog Identification Based on Multi-Satellite Information and the ECA-TransUnet Model
Sea fog is a weather hazard along the coast and over the ocean that seriously threatens maritime activities. In the deep learning approach, it is difficult for convolutional neural networks (CNNs) to fully consider global context information in sea fog research due to their own limitations, and the...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-08-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/15/16/3949 |
| _version_ | 1827728784656171008 |
|---|---|
| author | He Lu Yi Ma Shichao Zhang Xiang Yu Jiahua Zhang |
| author_facet | He Lu Yi Ma Shichao Zhang Xiang Yu Jiahua Zhang |
| author_sort | He Lu |
| collection | DOAJ |
| description | Sea fog is a weather hazard along the coast and over the ocean that seriously threatens maritime activities. In the deep learning approach, it is difficult for convolutional neural networks (CNNs) to fully consider global context information in sea fog research due to their own limitations, and the recognition of sea fog edges is relatively vague. To solve the above problems, this paper puts forward an ECA-TransUnet model for daytime sea fog recognition, which consists of a combination of a CNN and a transformer. By designing a two-branch feed-forward network (FFN) module and introducing an efficient channel attention (ECA) module, the model can effectively take into account long-range pixel interactions and feature channel information to capture the global contextual information of sea fog data. Meanwhile, to solve the problem of insufficient existing sea fog detection datasets, we investigated sea fog events occurring in the Yellow Sea and Bohai Sea and their territorial waters, extracted remote sensing images from Moderate Resolution Imaging Spectroradiometer (MODIS) data at corresponding times, and combined data from the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO), cloud and sea fog texture features, and waveband feature information to produce a manually annotated sea fog dataset. Our experiments showed that the proposed model achieves 94.5% accuracy and an 85.8% F1 score. Compared with the existing models relying only on CNNs such as UNet, FCN8s, and DeeplabV3+, it achieves state-of-the-art performance in sea fog recognition. |
| first_indexed | 2024-03-10T23:36:49Z |
| format | Article |
| id | doaj.art-919f51fdec8e44d497585cbedd469b1d |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-10T23:36:49Z |
| publishDate | 2023-08-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-919f51fdec8e44d497585cbedd469b1d2023-11-19T02:52:25ZengMDPI AGRemote Sensing2072-42922023-08-011516394910.3390/rs15163949Daytime Sea Fog Identification Based on Multi-Satellite Information and the ECA-TransUnet ModelHe Lu0Yi Ma1Shichao Zhang2Xiang Yu3Jiahua Zhang4Remote Sensing Information and Digital Earth Center, College of Computer Science and Technology, Qingdao University, Qingdao 266071, ChinaFirst Institute of Oceanography, State Oceanic Administration, Qingdao 266061, ChinaRemote Sensing Information and Digital Earth Center, College of Computer Science and Technology, Qingdao University, Qingdao 266071, ChinaRemote Sensing Information and Digital Earth Center, College of Computer Science and Technology, Qingdao University, Qingdao 266071, ChinaRemote Sensing Information and Digital Earth Center, College of Computer Science and Technology, Qingdao University, Qingdao 266071, ChinaSea fog is a weather hazard along the coast and over the ocean that seriously threatens maritime activities. In the deep learning approach, it is difficult for convolutional neural networks (CNNs) to fully consider global context information in sea fog research due to their own limitations, and the recognition of sea fog edges is relatively vague. To solve the above problems, this paper puts forward an ECA-TransUnet model for daytime sea fog recognition, which consists of a combination of a CNN and a transformer. By designing a two-branch feed-forward network (FFN) module and introducing an efficient channel attention (ECA) module, the model can effectively take into account long-range pixel interactions and feature channel information to capture the global contextual information of sea fog data. Meanwhile, to solve the problem of insufficient existing sea fog detection datasets, we investigated sea fog events occurring in the Yellow Sea and Bohai Sea and their territorial waters, extracted remote sensing images from Moderate Resolution Imaging Spectroradiometer (MODIS) data at corresponding times, and combined data from the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO), cloud and sea fog texture features, and waveband feature information to produce a manually annotated sea fog dataset. Our experiments showed that the proposed model achieves 94.5% accuracy and an 85.8% F1 score. Compared with the existing models relying only on CNNs such as UNet, FCN8s, and DeeplabV3+, it achieves state-of-the-art performance in sea fog recognition.https://www.mdpi.com/2072-4292/15/16/3949sea fog recognitiondeep learningtransformerconvolutional neural network (CNN)efficient channel attention (ECA) |
| spellingShingle | He Lu Yi Ma Shichao Zhang Xiang Yu Jiahua Zhang Daytime Sea Fog Identification Based on Multi-Satellite Information and the ECA-TransUnet Model Remote Sensing sea fog recognition deep learning transformer convolutional neural network (CNN) efficient channel attention (ECA) |
| title | Daytime Sea Fog Identification Based on Multi-Satellite Information and the ECA-TransUnet Model |
| title_full | Daytime Sea Fog Identification Based on Multi-Satellite Information and the ECA-TransUnet Model |
| title_fullStr | Daytime Sea Fog Identification Based on Multi-Satellite Information and the ECA-TransUnet Model |
| title_full_unstemmed | Daytime Sea Fog Identification Based on Multi-Satellite Information and the ECA-TransUnet Model |
| title_short | Daytime Sea Fog Identification Based on Multi-Satellite Information and the ECA-TransUnet Model |
| title_sort | daytime sea fog identification based on multi satellite information and the eca transunet model |
| topic | sea fog recognition deep learning transformer convolutional neural network (CNN) efficient channel attention (ECA) |
| url | https://www.mdpi.com/2072-4292/15/16/3949 |
| work_keys_str_mv | AT helu daytimeseafogidentificationbasedonmultisatelliteinformationandtheecatransunetmodel AT yima daytimeseafogidentificationbasedonmultisatelliteinformationandtheecatransunetmodel AT shichaozhang daytimeseafogidentificationbasedonmultisatelliteinformationandtheecatransunetmodel AT xiangyu daytimeseafogidentificationbasedonmultisatelliteinformationandtheecatransunetmodel AT jiahuazhang daytimeseafogidentificationbasedonmultisatelliteinformationandtheecatransunetmodel |