|
A Classification of Crown Flower using Feature Extraction and Machine Learning |
|---|---|
| รหัสดีโอไอ | |
| Creator | On-Uma Pramote |
| Title | A Classification of Crown Flower using Feature Extraction and Machine Learning |
| Contributor | Sasin Tiendee, Kewalin Khamnipat |
| Publisher | The Association of Council of IT Deans (CITT) |
| Publication Year | 2565 |
| Journal Title | Journal of Information Science and Technology |
| Journal Vol. | 12 |
| Journal No. | 2 |
| Page no. | 1-10 |
| Keyword | Classification, Feature Extraction, Machine Leaning, Crown Flower |
| URL Website | https://tci-thaijo.org/index.php/JIST |
| Website title | Journal of Information Science and Technology |
| ISSN | 2651-1053 |
| Abstract | This research presents an algorithm for image analysis dahlia classification that uses machine learning techniques to learn the distinctive features of dahlia images. The dahlia flowers used in this study were divided into four classes: (1) well-formed flowers, (2) fungal-infected flowers, (3) unequal petals, and (4) basal flowers. The dahlia flowers used in this research were divided into four classes: (1) flowers with good appearance, (2) flowers with fungal attachment, (3) flowers with unequal petal length, and (4) flowers at the base that are adjacent to each other. There are 200 flowers per class, a total of 800 flowers. The main features used in the research were three attributes: (1) percentage of dark pixels, (2) petal area (3) area between petals. Three techniques were used in this study: (1) Decision Trees, (2) Support Vector Machines, and (3) Deep Learning. In this study, 70% of the total data was used to create the classifier, and the remaining 30% was used for the classifier test. The classifier benchmarks with ten-fold cross-validation were precision 95.92, recall 95.90, accuracy 95.89, and f-measure 95.91. Test datasets were used to measure performance and found that decision trees were the most efficient learning machines. The results were precision 99.59 %, recall 99.58%, accuracy 99.58%, and f-measure rate 99.58%. |
