An Efficient Weed Detection Method Using Latent Diffusion Transformer for Enhanced Agricultural Image Analysis and Mobile Deployment
This paper presents an efficient weed detection method based on the latent diffusion transformer, aimed at enhancing the accuracy and applicability of agricultural image analysis. The experimental results demonstrate that the proposed model achieves a precision of 0.92, a recall of 0.89, an accuracy...
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2024-11-01
|
| Series: | Plants |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2223-7747/13/22/3192 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1850147285484175360 |
|---|---|
| author | Yuzhuo Cui Yingqiu Yang Yuqing Xia Yan Li Zhaoxi Feng Shiya Liu Guangqi Yuan Chunli Lv |
| author_facet | Yuzhuo Cui Yingqiu Yang Yuqing Xia Yan Li Zhaoxi Feng Shiya Liu Guangqi Yuan Chunli Lv |
| author_sort | Yuzhuo Cui |
| collection | DOAJ |
| description | This paper presents an efficient weed detection method based on the latent diffusion transformer, aimed at enhancing the accuracy and applicability of agricultural image analysis. The experimental results demonstrate that the proposed model achieves a precision of 0.92, a recall of 0.89, an accuracy of 0.91, a mean average precision (mAP) of 0.91, and an F1 score of 0.90, indicating its outstanding performance in complex scenarios. Additionally, ablation experiments reveal that the latent-space-based diffusion subnetwork outperforms traditional models, such as the the residual diffusion network, which has a precision of only 0.75. By combining latent space feature extraction with self-attention mechanisms, the constructed lightweight model can respond quickly on mobile devices, showcasing the significant potential of deep learning technologies in agricultural applications. Future research will focus on data diversity and model interpretability to further enhance the model’s adaptability and user trust. |
| format | Article |
| id | doaj-art-e5fd3aa4b29141dba3825f68e7b1561d |
| institution | OA Journals |
| issn | 2223-7747 |
| language | English |
| publishDate | 2024-11-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Plants |
| spelling | doaj-art-e5fd3aa4b29141dba3825f68e7b1561d2025-08-20T02:27:36ZengMDPI AGPlants2223-77472024-11-011322319210.3390/plants13223192An Efficient Weed Detection Method Using Latent Diffusion Transformer for Enhanced Agricultural Image Analysis and Mobile DeploymentYuzhuo Cui0Yingqiu Yang1Yuqing Xia2Yan Li3Zhaoxi Feng4Shiya Liu5Guangqi Yuan6Chunli Lv7College of Electrical and Information Engineering, China Agricultural University, Beijing 100083, ChinaCollege of Electrical and Information Engineering, China Agricultural University, Beijing 100083, ChinaCollege of Electrical and Information Engineering, China Agricultural University, Beijing 100083, ChinaCollege of Electrical and Information Engineering, China Agricultural University, Beijing 100083, ChinaCollege of Electrical and Information Engineering, China Agricultural University, Beijing 100083, ChinaCollege of Electrical and Information Engineering, China Agricultural University, Beijing 100083, ChinaCollege of Electrical and Information Engineering, China Agricultural University, Beijing 100083, ChinaCollege of Electrical and Information Engineering, China Agricultural University, Beijing 100083, ChinaThis paper presents an efficient weed detection method based on the latent diffusion transformer, aimed at enhancing the accuracy and applicability of agricultural image analysis. The experimental results demonstrate that the proposed model achieves a precision of 0.92, a recall of 0.89, an accuracy of 0.91, a mean average precision (mAP) of 0.91, and an F1 score of 0.90, indicating its outstanding performance in complex scenarios. Additionally, ablation experiments reveal that the latent-space-based diffusion subnetwork outperforms traditional models, such as the the residual diffusion network, which has a precision of only 0.75. By combining latent space feature extraction with self-attention mechanisms, the constructed lightweight model can respond quickly on mobile devices, showcasing the significant potential of deep learning technologies in agricultural applications. Future research will focus on data diversity and model interpretability to further enhance the model’s adaptability and user trust.https://www.mdpi.com/2223-7747/13/22/3192weed detectionagricultural image analysislatent diffusion transformerreal-time monitoringdeep learning |
| spellingShingle | Yuzhuo Cui Yingqiu Yang Yuqing Xia Yan Li Zhaoxi Feng Shiya Liu Guangqi Yuan Chunli Lv An Efficient Weed Detection Method Using Latent Diffusion Transformer for Enhanced Agricultural Image Analysis and Mobile Deployment Plants weed detection agricultural image analysis latent diffusion transformer real-time monitoring deep learning |
| title | An Efficient Weed Detection Method Using Latent Diffusion Transformer for Enhanced Agricultural Image Analysis and Mobile Deployment |
| title_full | An Efficient Weed Detection Method Using Latent Diffusion Transformer for Enhanced Agricultural Image Analysis and Mobile Deployment |
| title_fullStr | An Efficient Weed Detection Method Using Latent Diffusion Transformer for Enhanced Agricultural Image Analysis and Mobile Deployment |
| title_full_unstemmed | An Efficient Weed Detection Method Using Latent Diffusion Transformer for Enhanced Agricultural Image Analysis and Mobile Deployment |
| title_short | An Efficient Weed Detection Method Using Latent Diffusion Transformer for Enhanced Agricultural Image Analysis and Mobile Deployment |
| title_sort | efficient weed detection method using latent diffusion transformer for enhanced agricultural image analysis and mobile deployment |
| topic | weed detection agricultural image analysis latent diffusion transformer real-time monitoring deep learning |
| url | https://www.mdpi.com/2223-7747/13/22/3192 |
| work_keys_str_mv | AT yuzhuocui anefficientweeddetectionmethodusinglatentdiffusiontransformerforenhancedagriculturalimageanalysisandmobiledeployment AT yingqiuyang anefficientweeddetectionmethodusinglatentdiffusiontransformerforenhancedagriculturalimageanalysisandmobiledeployment AT yuqingxia anefficientweeddetectionmethodusinglatentdiffusiontransformerforenhancedagriculturalimageanalysisandmobiledeployment AT yanli anefficientweeddetectionmethodusinglatentdiffusiontransformerforenhancedagriculturalimageanalysisandmobiledeployment AT zhaoxifeng anefficientweeddetectionmethodusinglatentdiffusiontransformerforenhancedagriculturalimageanalysisandmobiledeployment AT shiyaliu anefficientweeddetectionmethodusinglatentdiffusiontransformerforenhancedagriculturalimageanalysisandmobiledeployment AT guangqiyuan anefficientweeddetectionmethodusinglatentdiffusiontransformerforenhancedagriculturalimageanalysisandmobiledeployment AT chunlilv anefficientweeddetectionmethodusinglatentdiffusiontransformerforenhancedagriculturalimageanalysisandmobiledeployment AT yuzhuocui efficientweeddetectionmethodusinglatentdiffusiontransformerforenhancedagriculturalimageanalysisandmobiledeployment AT yingqiuyang efficientweeddetectionmethodusinglatentdiffusiontransformerforenhancedagriculturalimageanalysisandmobiledeployment AT yuqingxia efficientweeddetectionmethodusinglatentdiffusiontransformerforenhancedagriculturalimageanalysisandmobiledeployment AT yanli efficientweeddetectionmethodusinglatentdiffusiontransformerforenhancedagriculturalimageanalysisandmobiledeployment AT zhaoxifeng efficientweeddetectionmethodusinglatentdiffusiontransformerforenhancedagriculturalimageanalysisandmobiledeployment AT shiyaliu efficientweeddetectionmethodusinglatentdiffusiontransformerforenhancedagriculturalimageanalysisandmobiledeployment AT guangqiyuan efficientweeddetectionmethodusinglatentdiffusiontransformerforenhancedagriculturalimageanalysisandmobiledeployment AT chunlilv efficientweeddetectionmethodusinglatentdiffusiontransformerforenhancedagriculturalimageanalysisandmobiledeployment |