Plant and Disease Recognition Based on PMF Pipeline Domain Adaptation Method: Using Bark Images as Meta-Dataset
Efficient image recognition is important in crop and forest management. However, it faces many challenges, such as the large number of plant species and diseases, the variability of plant appearance, and the scarcity of labeled data for training. To address this issue, we modified a SOTA Cross-Domai...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-09-01
|
| Series: | Plants |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2223-7747/12/18/3280 |
| _version_ | 1827724297419882496 |
|---|---|
| author | Zhelin Cui Kanglong Li Chunyan Kang Yi Wu Tao Li Mingyang Li |
| author_facet | Zhelin Cui Kanglong Li Chunyan Kang Yi Wu Tao Li Mingyang Li |
| author_sort | Zhelin Cui |
| collection | DOAJ |
| description | Efficient image recognition is important in crop and forest management. However, it faces many challenges, such as the large number of plant species and diseases, the variability of plant appearance, and the scarcity of labeled data for training. To address this issue, we modified a SOTA Cross-Domain Few-shot Learning (CDFSL) method based on prototypical networks and attention mechanisms. We employed attention mechanisms to perform feature extraction and prototype generation by focusing on the most relevant parts of the images, then used prototypical networks to learn the prototype of each category and classify new instances. Finally, we demonstrated the effectiveness of the modified CDFSL method on several plant and disease recognition datasets. The results showed that the modified pipeline was able to recognize several cross-domain datasets using generic representations, and achieved up to 96.95% and 94.07% classification accuracy on datasets with the same and different domains, respectively. In addition, we visualized the experimental results, demonstrating the model’s stable transfer capability between datasets and the model’s high visual correlation with plant and disease biological characteristics. Moreover, by extending the classes of different semantics within the training dataset, our model can be generalized to other domains, which implies broad applicability. |
| first_indexed | 2024-03-10T22:13:47Z |
| format | Article |
| id | doaj.art-de226ccb472641f08c93177cb22d8635 |
| institution | Directory Open Access Journal |
| issn | 2223-7747 |
| language | English |
| last_indexed | 2024-03-10T22:13:47Z |
| publishDate | 2023-09-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Plants |
| spelling | doaj.art-de226ccb472641f08c93177cb22d86352023-11-19T12:32:39ZengMDPI AGPlants2223-77472023-09-011218328010.3390/plants12183280Plant and Disease Recognition Based on PMF Pipeline Domain Adaptation Method: Using Bark Images as Meta-DatasetZhelin Cui0Kanglong Li1Chunyan Kang2Yi Wu3Tao Li4Mingyang Li5Co-Innovation Center for Sustainable Forestry in Southern China, Nanjing Forestry University, Nanjing 210037, ChinaCo-Innovation Center for Sustainable Forestry in Southern China, Nanjing Forestry University, Nanjing 210037, ChinaCo-Innovation Center for Sustainable Forestry in Southern China, Nanjing Forestry University, Nanjing 210037, ChinaCo-Innovation Center for Sustainable Forestry in Southern China, Nanjing Forestry University, Nanjing 210037, ChinaCo-Innovation Center for Sustainable Forestry in Southern China, Nanjing Forestry University, Nanjing 210037, ChinaCo-Innovation Center for Sustainable Forestry in Southern China, Nanjing Forestry University, Nanjing 210037, ChinaEfficient image recognition is important in crop and forest management. However, it faces many challenges, such as the large number of plant species and diseases, the variability of plant appearance, and the scarcity of labeled data for training. To address this issue, we modified a SOTA Cross-Domain Few-shot Learning (CDFSL) method based on prototypical networks and attention mechanisms. We employed attention mechanisms to perform feature extraction and prototype generation by focusing on the most relevant parts of the images, then used prototypical networks to learn the prototype of each category and classify new instances. Finally, we demonstrated the effectiveness of the modified CDFSL method on several plant and disease recognition datasets. The results showed that the modified pipeline was able to recognize several cross-domain datasets using generic representations, and achieved up to 96.95% and 94.07% classification accuracy on datasets with the same and different domains, respectively. In addition, we visualized the experimental results, demonstrating the model’s stable transfer capability between datasets and the model’s high visual correlation with plant and disease biological characteristics. Moreover, by extending the classes of different semantics within the training dataset, our model can be generalized to other domains, which implies broad applicability.https://www.mdpi.com/2223-7747/12/18/3280image classificationfew-shot learningtransfer learningbark images dataset |
| spellingShingle | Zhelin Cui Kanglong Li Chunyan Kang Yi Wu Tao Li Mingyang Li Plant and Disease Recognition Based on PMF Pipeline Domain Adaptation Method: Using Bark Images as Meta-Dataset Plants image classification few-shot learning transfer learning bark images dataset |
| title | Plant and Disease Recognition Based on PMF Pipeline Domain Adaptation Method: Using Bark Images as Meta-Dataset |
| title_full | Plant and Disease Recognition Based on PMF Pipeline Domain Adaptation Method: Using Bark Images as Meta-Dataset |
| title_fullStr | Plant and Disease Recognition Based on PMF Pipeline Domain Adaptation Method: Using Bark Images as Meta-Dataset |
| title_full_unstemmed | Plant and Disease Recognition Based on PMF Pipeline Domain Adaptation Method: Using Bark Images as Meta-Dataset |
| title_short | Plant and Disease Recognition Based on PMF Pipeline Domain Adaptation Method: Using Bark Images as Meta-Dataset |
| title_sort | plant and disease recognition based on pmf pipeline domain adaptation method using bark images as meta dataset |
| topic | image classification few-shot learning transfer learning bark images dataset |
| url | https://www.mdpi.com/2223-7747/12/18/3280 |
| work_keys_str_mv | AT zhelincui plantanddiseaserecognitionbasedonpmfpipelinedomainadaptationmethodusingbarkimagesasmetadataset AT kanglongli plantanddiseaserecognitionbasedonpmfpipelinedomainadaptationmethodusingbarkimagesasmetadataset AT chunyankang plantanddiseaserecognitionbasedonpmfpipelinedomainadaptationmethodusingbarkimagesasmetadataset AT yiwu plantanddiseaserecognitionbasedonpmfpipelinedomainadaptationmethodusingbarkimagesasmetadataset AT taoli plantanddiseaserecognitionbasedonpmfpipelinedomainadaptationmethodusingbarkimagesasmetadataset AT mingyangli plantanddiseaserecognitionbasedonpmfpipelinedomainadaptationmethodusingbarkimagesasmetadataset |