| Resumen |
Ethiopia is one of the African countries with a high potential for the creation of several crop varieties utilized in traditional medicine and daily life. Ginger is one of the plants that is afflicted by illness. The detection of disease requires specific attention from professionals, which is not achievable for mass production. On the other hand, cutting-edge technology can be used to circumvent the issue by applying image processing to a crop of ginger that is grown on a large scale. To that end, early ginger disease detection from the leaf is presented in stages after collecting 7,014 ginger photos with the assistance of domain experts from various farms using a transfer learning approach. The obtained data were subjected to various picture pre-processing techniques in order to construct and develop a model capable of detecting and dealing with a variety of circumstances. The study conducted two experiments: one using a pre-trained model for feature extraction, which has an accuracy of 91%, and the other fine-tuning a pre-trained model, which has superior performance than a pre-trained model for feature extraction, which has an accuracy of 97%. The experimental results show that the suggested technique is effective for detecting ginger diseases, particularly bacterial wilt. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG. |
