Lightweight Asymmetric Dilation Network for Real-Time Semantic Segmentation
Semantic segmentation is a very important and challenging problem in computer vision. Many applications, such as automated driving and robotic navigation in urban road scenes, require accurate and efficient segmentation. Nowadays, system models are often designed with high speed but a large number o...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2021-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9399082/ |
| _version_ | 1819121338577584128 |
|---|---|
| author | Xuegang Hu Yu Gong |
| author_facet | Xuegang Hu Yu Gong |
| author_sort | Xuegang Hu |
| collection | DOAJ |
| description | Semantic segmentation is a very important and challenging problem in computer vision. Many applications, such as automated driving and robotic navigation in urban road scenes, require accurate and efficient segmentation. Nowadays, system models are often designed with high speed but a large number of parameters, or they take up a lot of memory space with a very small speed, so they are not suitable for real-time semantic segmentation conditions. In order to solve this problem, we propose a more comprehensive model that has not only a faster speed, but also a smaller number of parameters and a higher accuracy which is termed as Lightweight Asymmetric Dilation Network (LADNet). Our model is based on our Lightweight Asymmetric Dilation Module (LAD Module) which provides a larger receptive field than all existing lightweight models to learn more information, while Lightweight Asymmetric Dilation-A (LAD-A) can better perceive spatial and semantic information, and Lightweight Asymmetric Dilation-B (LAD-B) can better perceive semantic information. Our Lightweight Downsampling Module (LDM) downsamples the feature map, it can greatly reduce model parameters. Finally, our Attention Enhancement Decoder (AED) to restore the feature map to the same size as the resolution of the original image, AED enables two attentional feature maps to simultaneously guide semantic information for better semantic segmentation of images. Our extensive experiments on the Cityscapes, CamVid, and NYUv2 test set show that our model is able to achieve the best balance in parameters, accuracy, and speed. |
| first_indexed | 2024-12-22T06:34:59Z |
| format | Article |
| id | doaj.art-ca9e138eb1814267b2c0102126221a5a |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-12-22T06:34:59Z |
| publishDate | 2021-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-ca9e138eb1814267b2c0102126221a5a2022-12-21T18:35:36ZengIEEEIEEE Access2169-35362021-01-019556305564310.1109/ACCESS.2021.30718669399082Lightweight Asymmetric Dilation Network for Real-Time Semantic SegmentationXuegang Hu0Yu Gong1https://orcid.org/0000-0003-2836-2890School of Communication and Information Engineering, Chongqing University of Posts and Telecommunications, Chongqing, ChinaLaboratory of Intelligent Analysis and Decision on Complex Systems, Chongqing University of Posts and Telecommunications, Chongqing, ChinaSemantic segmentation is a very important and challenging problem in computer vision. Many applications, such as automated driving and robotic navigation in urban road scenes, require accurate and efficient segmentation. Nowadays, system models are often designed with high speed but a large number of parameters, or they take up a lot of memory space with a very small speed, so they are not suitable for real-time semantic segmentation conditions. In order to solve this problem, we propose a more comprehensive model that has not only a faster speed, but also a smaller number of parameters and a higher accuracy which is termed as Lightweight Asymmetric Dilation Network (LADNet). Our model is based on our Lightweight Asymmetric Dilation Module (LAD Module) which provides a larger receptive field than all existing lightweight models to learn more information, while Lightweight Asymmetric Dilation-A (LAD-A) can better perceive spatial and semantic information, and Lightweight Asymmetric Dilation-B (LAD-B) can better perceive semantic information. Our Lightweight Downsampling Module (LDM) downsamples the feature map, it can greatly reduce model parameters. Finally, our Attention Enhancement Decoder (AED) to restore the feature map to the same size as the resolution of the original image, AED enables two attentional feature maps to simultaneously guide semantic information for better semantic segmentation of images. Our extensive experiments on the Cityscapes, CamVid, and NYUv2 test set show that our model is able to achieve the best balance in parameters, accuracy, and speed.https://ieeexplore.ieee.org/document/9399082/Attention mechanismconvolutional neural networkencoder-decoder networklightweight modelreal-time semantic segmentation |
| spellingShingle | Xuegang Hu Yu Gong Lightweight Asymmetric Dilation Network for Real-Time Semantic Segmentation IEEE Access Attention mechanism convolutional neural network encoder-decoder network lightweight model real-time semantic segmentation |
| title | Lightweight Asymmetric Dilation Network for Real-Time Semantic Segmentation |
| title_full | Lightweight Asymmetric Dilation Network for Real-Time Semantic Segmentation |
| title_fullStr | Lightweight Asymmetric Dilation Network for Real-Time Semantic Segmentation |
| title_full_unstemmed | Lightweight Asymmetric Dilation Network for Real-Time Semantic Segmentation |
| title_short | Lightweight Asymmetric Dilation Network for Real-Time Semantic Segmentation |
| title_sort | lightweight asymmetric dilation network for real time semantic segmentation |
| topic | Attention mechanism convolutional neural network encoder-decoder network lightweight model real-time semantic segmentation |
| url | https://ieeexplore.ieee.org/document/9399082/ |
| work_keys_str_mv | AT xueganghu lightweightasymmetricdilationnetworkforrealtimesemanticsegmentation AT yugong lightweightasymmetricdilationnetworkforrealtimesemanticsegmentation |