|
Pest Insect Classification from Photographic Images Using Pre-Trained Deep Learning Models: A Case Study in Ubon Ratchathani Province |
|---|---|
| รหัสดีโอไอ | |
| Creator | Wichit Sombat |
| Title | Pest Insect Classification from Photographic Images Using Pre-Trained Deep Learning Models: A Case Study in Ubon Ratchathani Province |
| Contributor | Phaichayon Kongchai |
| Publisher | Faculty of Science, Ubon Ratchathani University |
| Publication Year | 2568 |
| Journal Title | Journal of Science and Science Education |
| Journal Vol. | 8 |
| Journal No. | 1 |
| Page no. | 168-179 |
| Keyword | Deep Learning, Image Classification, Pre-trained Models, Pest Insects, Sustainable Agriculture |
| URL Website | https://so04.tci-thaijo.org/index.php/JSSE/ |
| Website title | Journal of Science and Science Education |
| ISSN | ISSN 2697-410X |
| Abstract | This research aims to compare the performance of pre-trained deep learning models, namely MobileNetV2, ResNet50, and VGG19, with traditional machine learning algorithms, including Random Forest, Support Vector Machine (SVM), and Logistic Regression (LR), in classifying images of 10 pest insect species that impact key economic crops in Ubon Ratchathani Province. A dataset consisting of 4,875 images was utilized for this study. The research methodology involved feature extraction using pre-trained deep learning models, followed by training traditional algorithms with the extracted features. The experiments were conducted under a supervised learning framework, with parameter settings kept constant to ensure fair comparisons. The results showed that feature extraction with MobileNetV2 combined with Logistic Regression achieved the best performance, yielding the highest accuracy of 0.898 and an F1-Score of 0.869. This study highlights the potential of integrating deep learning techniques with traditional algorithms as an effective approach for improving classification tasks. Such integration can play a pivotal role in enhancing agricultural sustainability and efficiency in the future. |
