Image-to-Image Translation-Based Data Augmentation for Improving Crop/Weed Classification Models for Precision Agriculture Applications
Applications of deep-learning models in machine visions for crop/weed identification have remarkably upgraded the authenticity of precise weed management. However, compelling data are required to obtain the desired result from this highly data-driven operation. This study aims to curtail the effort...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-10-01
|
| Series: | Algorithms |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1999-4893/15/11/401 |
| _version_ | 1797469387809095680 |
|---|---|
| author | L. G. Divyanth D. S. Guru Peeyush Soni Rajendra Machavaram Mohammad Nadimi Jitendra Paliwal |
| author_facet | L. G. Divyanth D. S. Guru Peeyush Soni Rajendra Machavaram Mohammad Nadimi Jitendra Paliwal |
| author_sort | L. G. Divyanth |
| collection | DOAJ |
| description | Applications of deep-learning models in machine visions for crop/weed identification have remarkably upgraded the authenticity of precise weed management. However, compelling data are required to obtain the desired result from this highly data-driven operation. This study aims to curtail the effort needed to prepare very large image datasets by creating artificial images of maize (<i>Zea mays</i>) and four common weeds (i.e., Charlock, Fat Hen, Shepherd’s Purse, and small-flowered Cranesbill) through conditional Generative Adversarial Networks (cGANs). The fidelity of these synthetic images was tested through t-distributed stochastic neighbor embedding (t-SNE) visualization plots of real and artificial images of each class. The reliability of this method as a data augmentation technique was validated through classification results based on the transfer learning of a pre-defined convolutional neural network (CNN) architecture—the <i>AlexNet</i>; the feature extraction method came from the deepest pooling layer of the same network. Machine learning models based on a support vector machine (SVM) and linear discriminant analysis (LDA) were trained using these feature vectors. The <i>F</i>1 scores of the transfer learning model increased from 0.97 to 0.99, when additionally supported by an artificial dataset. Similarly, in the case of the feature extraction technique, the classification <i>F</i>1-scores increased from 0.93 to 0.96 for SVM and from 0.94 to 0.96 for the LDA model. The results show that image augmentation using generative adversarial networks (GANs) can improve the performance of crop/weed classification models with the added advantage of reduced time and manpower. Furthermore, it has demonstrated that generative networks could be a great tool for deep-learning applications in agriculture. |
| first_indexed | 2024-03-09T19:20:41Z |
| format | Article |
| id | doaj.art-133348878b5f48fdbfc02fab7ad6b1f3 |
| institution | Directory Open Access Journal |
| issn | 1999-4893 |
| language | English |
| last_indexed | 2024-03-09T19:20:41Z |
| publishDate | 2022-10-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Algorithms |
| spelling | doaj.art-133348878b5f48fdbfc02fab7ad6b1f32023-11-24T03:22:54ZengMDPI AGAlgorithms1999-48932022-10-01151140110.3390/a15110401Image-to-Image Translation-Based Data Augmentation for Improving Crop/Weed Classification Models for Precision Agriculture ApplicationsL. G. Divyanth0D. S. Guru1Peeyush Soni2Rajendra Machavaram3Mohammad Nadimi4Jitendra Paliwal5Department of Agricultural and Food Engineering, Indian Institute of Technology Kharagpur, Kharagpur 721302, IndiaDepartment of Studies in Computer Science, University of Mysore, Mysore 570006, IndiaDepartment of Agricultural and Food Engineering, Indian Institute of Technology Kharagpur, Kharagpur 721302, IndiaDepartment of Agricultural and Food Engineering, Indian Institute of Technology Kharagpur, Kharagpur 721302, IndiaDepartment of Biosystems Engineering, University of Manitoba, Winnipeg, MB R3T 5V6, CanadaDepartment of Biosystems Engineering, University of Manitoba, Winnipeg, MB R3T 5V6, CanadaApplications of deep-learning models in machine visions for crop/weed identification have remarkably upgraded the authenticity of precise weed management. However, compelling data are required to obtain the desired result from this highly data-driven operation. This study aims to curtail the effort needed to prepare very large image datasets by creating artificial images of maize (<i>Zea mays</i>) and four common weeds (i.e., Charlock, Fat Hen, Shepherd’s Purse, and small-flowered Cranesbill) through conditional Generative Adversarial Networks (cGANs). The fidelity of these synthetic images was tested through t-distributed stochastic neighbor embedding (t-SNE) visualization plots of real and artificial images of each class. The reliability of this method as a data augmentation technique was validated through classification results based on the transfer learning of a pre-defined convolutional neural network (CNN) architecture—the <i>AlexNet</i>; the feature extraction method came from the deepest pooling layer of the same network. Machine learning models based on a support vector machine (SVM) and linear discriminant analysis (LDA) were trained using these feature vectors. The <i>F</i>1 scores of the transfer learning model increased from 0.97 to 0.99, when additionally supported by an artificial dataset. Similarly, in the case of the feature extraction technique, the classification <i>F</i>1-scores increased from 0.93 to 0.96 for SVM and from 0.94 to 0.96 for the LDA model. The results show that image augmentation using generative adversarial networks (GANs) can improve the performance of crop/weed classification models with the added advantage of reduced time and manpower. Furthermore, it has demonstrated that generative networks could be a great tool for deep-learning applications in agriculture.https://www.mdpi.com/1999-4893/15/11/401generative adversarial networksdeep-learningcrop/weed classificationtransfer learningfeature extraction |
| spellingShingle | L. G. Divyanth D. S. Guru Peeyush Soni Rajendra Machavaram Mohammad Nadimi Jitendra Paliwal Image-to-Image Translation-Based Data Augmentation for Improving Crop/Weed Classification Models for Precision Agriculture Applications Algorithms generative adversarial networks deep-learning crop/weed classification transfer learning feature extraction |
| title | Image-to-Image Translation-Based Data Augmentation for Improving Crop/Weed Classification Models for Precision Agriculture Applications |
| title_full | Image-to-Image Translation-Based Data Augmentation for Improving Crop/Weed Classification Models for Precision Agriculture Applications |
| title_fullStr | Image-to-Image Translation-Based Data Augmentation for Improving Crop/Weed Classification Models for Precision Agriculture Applications |
| title_full_unstemmed | Image-to-Image Translation-Based Data Augmentation for Improving Crop/Weed Classification Models for Precision Agriculture Applications |
| title_short | Image-to-Image Translation-Based Data Augmentation for Improving Crop/Weed Classification Models for Precision Agriculture Applications |
| title_sort | image to image translation based data augmentation for improving crop weed classification models for precision agriculture applications |
| topic | generative adversarial networks deep-learning crop/weed classification transfer learning feature extraction |
| url | https://www.mdpi.com/1999-4893/15/11/401 |
| work_keys_str_mv | AT lgdivyanth imagetoimagetranslationbaseddataaugmentationforimprovingcropweedclassificationmodelsforprecisionagricultureapplications AT dsguru imagetoimagetranslationbaseddataaugmentationforimprovingcropweedclassificationmodelsforprecisionagricultureapplications AT peeyushsoni imagetoimagetranslationbaseddataaugmentationforimprovingcropweedclassificationmodelsforprecisionagricultureapplications AT rajendramachavaram imagetoimagetranslationbaseddataaugmentationforimprovingcropweedclassificationmodelsforprecisionagricultureapplications AT mohammadnadimi imagetoimagetranslationbaseddataaugmentationforimprovingcropweedclassificationmodelsforprecisionagricultureapplications AT jitendrapaliwal imagetoimagetranslationbaseddataaugmentationforimprovingcropweedclassificationmodelsforprecisionagricultureapplications |