An improved ShuffleNetV2 method based on ensemble self-distillation for tomato leaf diseases recognition
IntroductionTimely and accurate recognition of tomato diseases is crucial for improving tomato yield. While large deep learning models can achieve high-precision disease recognition, these models often have a large number of parameters, making them difficult to deploy on edge devices. To address thi...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2025-01-01
|
| Series: | Frontiers in Plant Science |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fpls.2024.1521008/full |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832592406602579968 |
|---|---|
| author | Shuiping Ni Yue Jia Mingfu Zhu Mingfu Zhu Yizhe Zhang Wendi Wang Shangxin Liu Yawei Chen |
| author_facet | Shuiping Ni Yue Jia Mingfu Zhu Mingfu Zhu Yizhe Zhang Wendi Wang Shangxin Liu Yawei Chen |
| author_sort | Shuiping Ni |
| collection | DOAJ |
| description | IntroductionTimely and accurate recognition of tomato diseases is crucial for improving tomato yield. While large deep learning models can achieve high-precision disease recognition, these models often have a large number of parameters, making them difficult to deploy on edge devices. To address this issue, this study proposes an ensemble self-distillation method and applies it to the lightweight model ShuffleNetV2.MethodsSpecifically, based on the architecture of ShuffleNetV2, multiple shallow models at different depths are constructed to establish a distillation framework. Based on the fused feature map that integrates the intermediate feature maps of ShuffleNetV2 and shallow models, a depthwise separable convolution layer is introduced to further extract more effective feature information. This method ensures that the intermediate features from each model are fully preserved to the ensemble model, thereby improving the overall performance of the ensemble model. The ensemble model, acting as the teacher, dynamically transfers knowledge to ShuffleNetV2 and the shallow models during training, significantly enhancing the performance of ShuffleNetV2 without changing the original structure.ResultsExperimental results show that the optimized ShuffleNetV2 achieves an accuracy of 95.08%, precision of 94.58%, recall of 94.55%, and an F1 score of 94.54% on the test set, surpassing large models such as VGG16 and ResNet18. Among lightweight models, it has the smallest parameter count and the highest recognition accuracy.DiscussionThe results demonstrate that the optimized ShuffleNetV2 is more suitable for deployment on edge devices for real-time tomato disease detection. Additionally, multiple shallow models achieve varying degrees of compression for ShuffleNetV2, providing flexibility for model deployment. |
| format | Article |
| id | doaj-art-409f3b1ece3d41428b3b44e7d1a08c90 |
| institution | Kabale University |
| issn | 1664-462X |
| language | English |
| publishDate | 2025-01-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Plant Science |
| spelling | doaj-art-409f3b1ece3d41428b3b44e7d1a08c902025-01-21T08:36:46ZengFrontiers Media S.A.Frontiers in Plant Science1664-462X2025-01-011510.3389/fpls.2024.15210081521008An improved ShuffleNetV2 method based on ensemble self-distillation for tomato leaf diseases recognitionShuiping Ni0Yue Jia1Mingfu Zhu2Mingfu Zhu3Yizhe Zhang4Wendi Wang5Shangxin Liu6Yawei Chen7School of Computer Science and Technology, Henan Polytechnic University, Jiaozuo, ChinaSchool of Computer Science and Technology, Henan Polytechnic University, Jiaozuo, ChinaSchool of Computer Science and Technology, Henan Polytechnic University, Jiaozuo, ChinaResearch and Development Department, Henan Chuitian Technology Corporation Limited, Hebi, ChinaSchool of Computer Science and Technology, Henan Polytechnic University, Jiaozuo, ChinaSchool of Computer Science and Technology, Henan Polytechnic University, Jiaozuo, ChinaResearch and Development Department, Henan Chuitian Technology Corporation Limited, Hebi, ChinaResearch and Development Department, Henan Chuitian Technology Corporation Limited, Hebi, ChinaIntroductionTimely and accurate recognition of tomato diseases is crucial for improving tomato yield. While large deep learning models can achieve high-precision disease recognition, these models often have a large number of parameters, making them difficult to deploy on edge devices. To address this issue, this study proposes an ensemble self-distillation method and applies it to the lightweight model ShuffleNetV2.MethodsSpecifically, based on the architecture of ShuffleNetV2, multiple shallow models at different depths are constructed to establish a distillation framework. Based on the fused feature map that integrates the intermediate feature maps of ShuffleNetV2 and shallow models, a depthwise separable convolution layer is introduced to further extract more effective feature information. This method ensures that the intermediate features from each model are fully preserved to the ensemble model, thereby improving the overall performance of the ensemble model. The ensemble model, acting as the teacher, dynamically transfers knowledge to ShuffleNetV2 and the shallow models during training, significantly enhancing the performance of ShuffleNetV2 without changing the original structure.ResultsExperimental results show that the optimized ShuffleNetV2 achieves an accuracy of 95.08%, precision of 94.58%, recall of 94.55%, and an F1 score of 94.54% on the test set, surpassing large models such as VGG16 and ResNet18. Among lightweight models, it has the smallest parameter count and the highest recognition accuracy.DiscussionThe results demonstrate that the optimized ShuffleNetV2 is more suitable for deployment on edge devices for real-time tomato disease detection. Additionally, multiple shallow models achieve varying degrees of compression for ShuffleNetV2, providing flexibility for model deployment.https://www.frontiersin.org/articles/10.3389/fpls.2024.1521008/fulltomato leaf diseases recognitionlightweight modelShuffleNetV2ensembleself-distillationmodel compression |
| spellingShingle | Shuiping Ni Yue Jia Mingfu Zhu Mingfu Zhu Yizhe Zhang Wendi Wang Shangxin Liu Yawei Chen An improved ShuffleNetV2 method based on ensemble self-distillation for tomato leaf diseases recognition Frontiers in Plant Science tomato leaf diseases recognition lightweight model ShuffleNetV2 ensemble self-distillation model compression |
| title | An improved ShuffleNetV2 method based on ensemble self-distillation for tomato leaf diseases recognition |
| title_full | An improved ShuffleNetV2 method based on ensemble self-distillation for tomato leaf diseases recognition |
| title_fullStr | An improved ShuffleNetV2 method based on ensemble self-distillation for tomato leaf diseases recognition |
| title_full_unstemmed | An improved ShuffleNetV2 method based on ensemble self-distillation for tomato leaf diseases recognition |
| title_short | An improved ShuffleNetV2 method based on ensemble self-distillation for tomato leaf diseases recognition |
| title_sort | improved shufflenetv2 method based on ensemble self distillation for tomato leaf diseases recognition |
| topic | tomato leaf diseases recognition lightweight model ShuffleNetV2 ensemble self-distillation model compression |
| url | https://www.frontiersin.org/articles/10.3389/fpls.2024.1521008/full |
| work_keys_str_mv | AT shuipingni animprovedshufflenetv2methodbasedonensembleselfdistillationfortomatoleafdiseasesrecognition AT yuejia animprovedshufflenetv2methodbasedonensembleselfdistillationfortomatoleafdiseasesrecognition AT mingfuzhu animprovedshufflenetv2methodbasedonensembleselfdistillationfortomatoleafdiseasesrecognition AT mingfuzhu animprovedshufflenetv2methodbasedonensembleselfdistillationfortomatoleafdiseasesrecognition AT yizhezhang animprovedshufflenetv2methodbasedonensembleselfdistillationfortomatoleafdiseasesrecognition AT wendiwang animprovedshufflenetv2methodbasedonensembleselfdistillationfortomatoleafdiseasesrecognition AT shangxinliu animprovedshufflenetv2methodbasedonensembleselfdistillationfortomatoleafdiseasesrecognition AT yaweichen animprovedshufflenetv2methodbasedonensembleselfdistillationfortomatoleafdiseasesrecognition AT shuipingni improvedshufflenetv2methodbasedonensembleselfdistillationfortomatoleafdiseasesrecognition AT yuejia improvedshufflenetv2methodbasedonensembleselfdistillationfortomatoleafdiseasesrecognition AT mingfuzhu improvedshufflenetv2methodbasedonensembleselfdistillationfortomatoleafdiseasesrecognition AT mingfuzhu improvedshufflenetv2methodbasedonensembleselfdistillationfortomatoleafdiseasesrecognition AT yizhezhang improvedshufflenetv2methodbasedonensembleselfdistillationfortomatoleafdiseasesrecognition AT wendiwang improvedshufflenetv2methodbasedonensembleselfdistillationfortomatoleafdiseasesrecognition AT shangxinliu improvedshufflenetv2methodbasedonensembleselfdistillationfortomatoleafdiseasesrecognition AT yaweichen improvedshufflenetv2methodbasedonensembleselfdistillationfortomatoleafdiseasesrecognition |