A Depth-Wise Separable U-Net Architecture with Multiscale Filters to Detect Sinkholes
Numerous variants of the basic deep segmentation model—U-Net—have emerged in recent years, achieving reliable performance across different benchmarks. In this paper, we propose an improved version of U-Net with higher performance and reduced complexity. This improvement was achieved by introducing a...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-02-01
|
| Series: | Remote Sensing |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2072-4292/15/5/1384 |
| _version_ | 1797614470570180608 |
|---|---|
| author | Rasha Alshawi Md Tamjidul Hoque Maik C. Flanagin |
| author_facet | Rasha Alshawi Md Tamjidul Hoque Maik C. Flanagin |
| author_sort | Rasha Alshawi |
| collection | DOAJ |
| description | Numerous variants of the basic deep segmentation model—U-Net—have emerged in recent years, achieving reliable performance across different benchmarks. In this paper, we propose an improved version of U-Net with higher performance and reduced complexity. This improvement was achieved by introducing a sparsely connected depth-wise separable block with multiscale filters, enabling the network to capture features of different scales. The use of depth-wise separable convolution significantly reduces the number of trainable parameters, making the training faster, while reducing the risk of overfitting. We used our developed sinkhole dataset and the available benchmark nuclei dataset to assess the proposed model’s performance. Pixel-wise annotation is laborious and requires a great deal of human expertise; therefore, we propose a fully deep convolutional autoencoder network that utilizes the proposed block to automatically annotate the sinkhole dataset. Our segmentation model outperformed the state-of-the-art methods, including U-Net, Attention U-Net, Depth-Separable U-Net, and Inception U-Net, achieving an average improvement of 1.2% and 1.4%, respectively, on the sinkhole and the nuclei datasets, with 94% and 92% accuracy, as well as a reduced training time. It also achieved 83% and 80% intersection-over-union (IoU) on the two datasets, respectively, which is an 11.8% and 9.3% average improvement over the above-mentioned models. |
| first_indexed | 2024-03-11T07:11:52Z |
| format | Article |
| id | doaj.art-c375722c7f5d455a96256357ebff3238 |
| institution | Directory Open Access Journal |
| issn | 2072-4292 |
| language | English |
| last_indexed | 2024-03-11T07:11:52Z |
| publishDate | 2023-02-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Remote Sensing |
| spelling | doaj.art-c375722c7f5d455a96256357ebff32382023-11-17T08:32:22ZengMDPI AGRemote Sensing2072-42922023-02-01155138410.3390/rs15051384A Depth-Wise Separable U-Net Architecture with Multiscale Filters to Detect SinkholesRasha Alshawi0Md Tamjidul Hoque1Maik C. Flanagin2Department of Computer Science, University of New Orleans, New Orleans, LA 70148, USADepartment of Computer Science, University of New Orleans, New Orleans, LA 70148, USAUS Army Corps of Engineers, New Orleans, LA 70118, USANumerous variants of the basic deep segmentation model—U-Net—have emerged in recent years, achieving reliable performance across different benchmarks. In this paper, we propose an improved version of U-Net with higher performance and reduced complexity. This improvement was achieved by introducing a sparsely connected depth-wise separable block with multiscale filters, enabling the network to capture features of different scales. The use of depth-wise separable convolution significantly reduces the number of trainable parameters, making the training faster, while reducing the risk of overfitting. We used our developed sinkhole dataset and the available benchmark nuclei dataset to assess the proposed model’s performance. Pixel-wise annotation is laborious and requires a great deal of human expertise; therefore, we propose a fully deep convolutional autoencoder network that utilizes the proposed block to automatically annotate the sinkhole dataset. Our segmentation model outperformed the state-of-the-art methods, including U-Net, Attention U-Net, Depth-Separable U-Net, and Inception U-Net, achieving an average improvement of 1.2% and 1.4%, respectively, on the sinkhole and the nuclei datasets, with 94% and 92% accuracy, as well as a reduced training time. It also achieved 83% and 80% intersection-over-union (IoU) on the two datasets, respectively, which is an 11.8% and 9.3% average improvement over the above-mentioned models.https://www.mdpi.com/2072-4292/15/5/1384sinkholedeep learningsemantic segmentationU-Netdepth-wise separable convolutionmultiscale filter |
| spellingShingle | Rasha Alshawi Md Tamjidul Hoque Maik C. Flanagin A Depth-Wise Separable U-Net Architecture with Multiscale Filters to Detect Sinkholes Remote Sensing sinkhole deep learning semantic segmentation U-Net depth-wise separable convolution multiscale filter |
| title | A Depth-Wise Separable U-Net Architecture with Multiscale Filters to Detect Sinkholes |
| title_full | A Depth-Wise Separable U-Net Architecture with Multiscale Filters to Detect Sinkholes |
| title_fullStr | A Depth-Wise Separable U-Net Architecture with Multiscale Filters to Detect Sinkholes |
| title_full_unstemmed | A Depth-Wise Separable U-Net Architecture with Multiscale Filters to Detect Sinkholes |
| title_short | A Depth-Wise Separable U-Net Architecture with Multiscale Filters to Detect Sinkholes |
| title_sort | depth wise separable u net architecture with multiscale filters to detect sinkholes |
| topic | sinkhole deep learning semantic segmentation U-Net depth-wise separable convolution multiscale filter |
| url | https://www.mdpi.com/2072-4292/15/5/1384 |
| work_keys_str_mv | AT rashaalshawi adepthwiseseparableunetarchitecturewithmultiscalefilterstodetectsinkholes AT mdtamjidulhoque adepthwiseseparableunetarchitecturewithmultiscalefilterstodetectsinkholes AT maikcflanagin adepthwiseseparableunetarchitecturewithmultiscalefilterstodetectsinkholes AT rashaalshawi depthwiseseparableunetarchitecturewithmultiscalefilterstodetectsinkholes AT mdtamjidulhoque depthwiseseparableunetarchitecturewithmultiscalefilterstodetectsinkholes AT maikcflanagin depthwiseseparableunetarchitecturewithmultiscalefilterstodetectsinkholes |