A novel multiscale feature enhancement network using learnable density map for red clustered pepper yield estimation
IntroductionAccurate and automated yield estimation for red cluster pepper (RCP) is essential to optimise field management and resource allocation. Traditional object detection-based methods for yield estimation often suffer from time-consuming and labour-intensive annotation processes, as well as s...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2025-04-01
|
| Series: | Frontiers in Plant Science |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fpls.2025.1548035/full |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850151575262068736 |
|---|---|
| author | Chenming Cheng Chenming Cheng Jin Lei Jin Lei Zicui Zhu Lijian Lu Lijian Lu Zhi Wang Zhi Wang Jiali Tao Jiali Tao Xinyan Qin Xinyan Qin |
| author_facet | Chenming Cheng Chenming Cheng Jin Lei Jin Lei Zicui Zhu Lijian Lu Lijian Lu Zhi Wang Zhi Wang Jiali Tao Jiali Tao Xinyan Qin Xinyan Qin |
| author_sort | Chenming Cheng |
| collection | DOAJ |
| description | IntroductionAccurate and automated yield estimation for red cluster pepper (RCP) is essential to optimise field management and resource allocation. Traditional object detection-based methods for yield estimation often suffer from time-consuming and labour-intensive annotation processes, as well as suboptimal accuracy in dense environments. To address these challenges, this paper proposes a novel multiscale feature enhancement network (MFEN) that integrates a learnable density map (LDM) for accurate RCP yield estimation.MethodsThe proposed method mainly involves three key steps. First, the kernel-based density map (KDM) method was improved by integrating the Swin Transformer (ST), resulting in LDM method, which produces higher quality density maps. Then, a novel MFEN was developed to improve feature extraction from these density maps. This network combines dilation convolution, residual structures, and an attention mechanism to effectively extract features. Finally, the LDM and the MFEN were jointly trained to estimate both yield and density maps for RCP.Results and discussionThe model achieved superior accuracy in RCP yield estimation by using LDM in conjunction with MFEN for joint training. Firstly, the integration of LDM significantly improved the accuracy of the model, with a 0.98% improvement over the previous iteration. Compared to other feature extraction networks, MFEN had the lowest mean absolute error (MAE) of 5.42, root mean square error (RMSE) of 10.37 and symmetric mean absolute percentage error (SMAPE) of 11.64%. It also achieved the highest R-squared (R²) value of 0.9802 on the test dataset, beating the best performing DSNet by 0.98%. Notably, despite its multi-column structure, the model has a significant advantage in terms of parameters, with only 13.08M parameters (a reduction of 3.18M compared to the classic single-column network CSRNet). This highlights the model’s ability to achieve the highest accuracy while maintaining efficient deployment capabilities. The proposed method provides an robust algorithmic support for efficient and intelligent yield estimation in RCP. |
| format | Article |
| id | doaj-art-cbf40c93d7db4e3bae5e36ef4cdb93a9 |
| institution | OA Journals |
| issn | 1664-462X |
| language | English |
| publishDate | 2025-04-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Plant Science |
| spelling | doaj-art-cbf40c93d7db4e3bae5e36ef4cdb93a92025-08-20T02:26:10ZengFrontiers Media S.A.Frontiers in Plant Science1664-462X2025-04-011610.3389/fpls.2025.15480351548035A novel multiscale feature enhancement network using learnable density map for red clustered pepper yield estimationChenming Cheng0Chenming Cheng1Jin Lei2Jin Lei3Zicui Zhu4Lijian Lu5Lijian Lu6Zhi Wang7Zhi Wang8Jiali Tao9Jiali Tao10Xinyan Qin11Xinyan Qin12College of Mechanical and Electrical Engineering, Shihezi University, Shihezi, ChinaEngineering Research Center for Production Mechanization of Oasis Characteristic Cash Crop, Ministry of Education, Shihezi, ChinaCollege of Mechanical and Electrical Engineering, Shihezi University, Shihezi, ChinaEngineering Research Center for Production Mechanization of Oasis Characteristic Cash Crop, Ministry of Education, Shihezi, ChinaTown Construction and Management Centre, 170th Regiment, 9th Division, Xinjiang Production and Construction Corps, Emin, ChinaCollege of Mechanical and Electrical Engineering, Shihezi University, Shihezi, ChinaKey Laboratory of Northwest Agricultural Equipment, Ministry of Agriculture and Rural Affairs, Shihezi, ChinaCollege of Mechanical and Electrical Engineering, Shihezi University, Shihezi, ChinaKey Laboratory of Northwest Agricultural Equipment, Ministry of Agriculture and Rural Affairs, Shihezi, ChinaCollege of Mechanical and Electrical Engineering, Shihezi University, Shihezi, ChinaKey Laboratory of Northwest Agricultural Equipment, Ministry of Agriculture and Rural Affairs, Shihezi, ChinaCollege of Mechanical and Electrical Engineering, Shihezi University, Shihezi, ChinaEngineering Research Center for Production Mechanization of Oasis Characteristic Cash Crop, Ministry of Education, Shihezi, ChinaIntroductionAccurate and automated yield estimation for red cluster pepper (RCP) is essential to optimise field management and resource allocation. Traditional object detection-based methods for yield estimation often suffer from time-consuming and labour-intensive annotation processes, as well as suboptimal accuracy in dense environments. To address these challenges, this paper proposes a novel multiscale feature enhancement network (MFEN) that integrates a learnable density map (LDM) for accurate RCP yield estimation.MethodsThe proposed method mainly involves three key steps. First, the kernel-based density map (KDM) method was improved by integrating the Swin Transformer (ST), resulting in LDM method, which produces higher quality density maps. Then, a novel MFEN was developed to improve feature extraction from these density maps. This network combines dilation convolution, residual structures, and an attention mechanism to effectively extract features. Finally, the LDM and the MFEN were jointly trained to estimate both yield and density maps for RCP.Results and discussionThe model achieved superior accuracy in RCP yield estimation by using LDM in conjunction with MFEN for joint training. Firstly, the integration of LDM significantly improved the accuracy of the model, with a 0.98% improvement over the previous iteration. Compared to other feature extraction networks, MFEN had the lowest mean absolute error (MAE) of 5.42, root mean square error (RMSE) of 10.37 and symmetric mean absolute percentage error (SMAPE) of 11.64%. It also achieved the highest R-squared (R²) value of 0.9802 on the test dataset, beating the best performing DSNet by 0.98%. Notably, despite its multi-column structure, the model has a significant advantage in terms of parameters, with only 13.08M parameters (a reduction of 3.18M compared to the classic single-column network CSRNet). This highlights the model’s ability to achieve the highest accuracy while maintaining efficient deployment capabilities. The proposed method provides an robust algorithmic support for efficient and intelligent yield estimation in RCP.https://www.frontiersin.org/articles/10.3389/fpls.2025.1548035/fullred clustered pepperyield estimationdensity map generationSwin Transformerhybrid dilation convolutionmultiscale feature enhancement network |
| spellingShingle | Chenming Cheng Chenming Cheng Jin Lei Jin Lei Zicui Zhu Lijian Lu Lijian Lu Zhi Wang Zhi Wang Jiali Tao Jiali Tao Xinyan Qin Xinyan Qin A novel multiscale feature enhancement network using learnable density map for red clustered pepper yield estimation Frontiers in Plant Science red clustered pepper yield estimation density map generation Swin Transformer hybrid dilation convolution multiscale feature enhancement network |
| title | A novel multiscale feature enhancement network using learnable density map for red clustered pepper yield estimation |
| title_full | A novel multiscale feature enhancement network using learnable density map for red clustered pepper yield estimation |
| title_fullStr | A novel multiscale feature enhancement network using learnable density map for red clustered pepper yield estimation |
| title_full_unstemmed | A novel multiscale feature enhancement network using learnable density map for red clustered pepper yield estimation |
| title_short | A novel multiscale feature enhancement network using learnable density map for red clustered pepper yield estimation |
| title_sort | novel multiscale feature enhancement network using learnable density map for red clustered pepper yield estimation |
| topic | red clustered pepper yield estimation density map generation Swin Transformer hybrid dilation convolution multiscale feature enhancement network |
| url | https://www.frontiersin.org/articles/10.3389/fpls.2025.1548035/full |
| work_keys_str_mv | AT chenmingcheng anovelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT chenmingcheng anovelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT jinlei anovelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT jinlei anovelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT zicuizhu anovelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT lijianlu anovelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT lijianlu anovelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT zhiwang anovelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT zhiwang anovelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT jialitao anovelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT jialitao anovelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT xinyanqin anovelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT xinyanqin anovelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT chenmingcheng novelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT chenmingcheng novelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT jinlei novelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT jinlei novelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT zicuizhu novelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT lijianlu novelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT lijianlu novelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT zhiwang novelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT zhiwang novelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT jialitao novelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT jialitao novelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT xinyanqin novelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation AT xinyanqin novelmultiscalefeatureenhancementnetworkusinglearnabledensitymapforredclusteredpepperyieldestimation |