PD-SegNet: Semantic Segmentation of Small Agricultural Targets in Complex Environments
The agricultural scene is a typical unstructured scene, which is intricate and heavily affected by sunlight, weather, and other factors. Agricultural segmentation targets are generally small in size and heavily obstructed. At the same time, image segmentation in agricultural scenes has strong applic...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2023-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10146268/ |
| _version_ | 1797692904369553408 |
|---|---|
| author | Zhijia Zhu Mingkun Jiang Jun Dong Shuang Wu Fan Ma |
| author_facet | Zhijia Zhu Mingkun Jiang Jun Dong Shuang Wu Fan Ma |
| author_sort | Zhijia Zhu |
| collection | DOAJ |
| description | The agricultural scene is a typical unstructured scene, which is intricate and heavily affected by sunlight, weather, and other factors. Agricultural segmentation targets are generally small in size and heavily obstructed. At the same time, image segmentation in agricultural scenes has strong application scenarios, such as blooming intensity estimation, which refers to the estimation of the density and intensity of blooms, fruit yield estimation, fruit harvesting positioning, and so on. Currently, CNNs dominate semantic segmentation of agricultural scenes due to the significant computational constraints of using the Transformer module. However, CNNs have several disadvantages, such as limited effective receptive fields and the inability to capture global information, which significantly reduce their segmentation accuracy in complex agricultural scenes. In addition, the simple upsampling process used in CNNs can result in blurred segmentation edges and inferior performance. This paper presents a new semantic segmentation algorithm based on SegFormer: PD-SegNet(Powerful Decoder SegFormer Network), which balances accuracy and computational efficiency and combines dynamic kernel self-renewal with edge-aware optimization. The proposed algorithm demonstrates outstanding performance in two typical agricultural scenarios: apple blossom and apple fruit segmentation detection and sets a new state-of-the-art (SOTA) on the MinneApple Apple segmentation dataset. Experimental results demonstrate that the proposed method outperforms the baseline method in the segmentation of complex small targets. This algorithm can optimize the semantics segmentation of small targets in complex scenes and contributes to the development of smart agriculture. |
| first_indexed | 2024-03-12T02:35:51Z |
| format | Article |
| id | doaj.art-e9966d632e2643abbd54ce1a56b92bdb |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-03-12T02:35:51Z |
| publishDate | 2023-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-e9966d632e2643abbd54ce1a56b92bdb2023-09-04T23:01:50ZengIEEEIEEE Access2169-35362023-01-0111902149022610.1109/ACCESS.2023.328403610146268PD-SegNet: Semantic Segmentation of Small Agricultural Targets in Complex EnvironmentsZhijia Zhu0https://orcid.org/0000-0003-2631-7551Mingkun Jiang1Jun Dong2Shuang Wu3Fan Ma4Science Island Branch, Graduate School of USTC, Hefei, ChinaHefei Institutes of Physical Science, Chinese Academy of Sciences, Hefei, ChinaHefei Institutes of Physical Science, Chinese Academy of Sciences, Hefei, ChinaHefei Institutes of Physical Science, Chinese Academy of Sciences, Hefei, ChinaHefei Institutes of Physical Science, Chinese Academy of Sciences, Hefei, ChinaThe agricultural scene is a typical unstructured scene, which is intricate and heavily affected by sunlight, weather, and other factors. Agricultural segmentation targets are generally small in size and heavily obstructed. At the same time, image segmentation in agricultural scenes has strong application scenarios, such as blooming intensity estimation, which refers to the estimation of the density and intensity of blooms, fruit yield estimation, fruit harvesting positioning, and so on. Currently, CNNs dominate semantic segmentation of agricultural scenes due to the significant computational constraints of using the Transformer module. However, CNNs have several disadvantages, such as limited effective receptive fields and the inability to capture global information, which significantly reduce their segmentation accuracy in complex agricultural scenes. In addition, the simple upsampling process used in CNNs can result in blurred segmentation edges and inferior performance. This paper presents a new semantic segmentation algorithm based on SegFormer: PD-SegNet(Powerful Decoder SegFormer Network), which balances accuracy and computational efficiency and combines dynamic kernel self-renewal with edge-aware optimization. The proposed algorithm demonstrates outstanding performance in two typical agricultural scenarios: apple blossom and apple fruit segmentation detection and sets a new state-of-the-art (SOTA) on the MinneApple Apple segmentation dataset. Experimental results demonstrate that the proposed method outperforms the baseline method in the segmentation of complex small targets. This algorithm can optimize the semantics segmentation of small targets in complex scenes and contributes to the development of smart agriculture.https://ieeexplore.ieee.org/document/10146268/Smart agriculturedeep learningsemantic segmentationobject detectiontransformerdynamic kernel |
| spellingShingle | Zhijia Zhu Mingkun Jiang Jun Dong Shuang Wu Fan Ma PD-SegNet: Semantic Segmentation of Small Agricultural Targets in Complex Environments IEEE Access Smart agriculture deep learning semantic segmentation object detection transformer dynamic kernel |
| title | PD-SegNet: Semantic Segmentation of Small Agricultural Targets in Complex Environments |
| title_full | PD-SegNet: Semantic Segmentation of Small Agricultural Targets in Complex Environments |
| title_fullStr | PD-SegNet: Semantic Segmentation of Small Agricultural Targets in Complex Environments |
| title_full_unstemmed | PD-SegNet: Semantic Segmentation of Small Agricultural Targets in Complex Environments |
| title_short | PD-SegNet: Semantic Segmentation of Small Agricultural Targets in Complex Environments |
| title_sort | pd segnet semantic segmentation of small agricultural targets in complex environments |
| topic | Smart agriculture deep learning semantic segmentation object detection transformer dynamic kernel |
| url | https://ieeexplore.ieee.org/document/10146268/ |
| work_keys_str_mv | AT zhijiazhu pdsegnetsemanticsegmentationofsmallagriculturaltargetsincomplexenvironments AT mingkunjiang pdsegnetsemanticsegmentationofsmallagriculturaltargetsincomplexenvironments AT jundong pdsegnetsemanticsegmentationofsmallagriculturaltargetsincomplexenvironments AT shuangwu pdsegnetsemanticsegmentationofsmallagriculturaltargetsincomplexenvironments AT fanma pdsegnetsemanticsegmentationofsmallagriculturaltargetsincomplexenvironments |