| Abstrakt: |
The importance of agriculture in a growing economy cannot be overstated, as it serves as the fundamental economic cornerstone of any nation. The sustainability of agriculture is threatened by plant viruses, leading to substantial financial losses. An effective solution is found in automatic plant disease detection methods, which streamline the monitoring process in large agricultural farms. Early detection of tomato and potato plant diseases involves a systematic approach, starting with the identification of affected plant parts, the observation of discolorations such as brown or black patches, and the search for signs of infesting insects. This research leverages machine learning and deep learning techniques to detect leaf diseases in tomato and potato plants. The datasets undergo thorough data augmentation and pre-processing methodologies before the application of machine learning and deep learning algorithms. In this study, we evaluate the performance of two prominent algorithms, Support Vector Machine (SVM) and Residual Network (ResNet). The comparison of simulation results aids in assessing the model's effectiveness. The input image analysis facilitates the detection of disease presence following the training process. Once tomato and potato plant leaf diseases are identified, we provide recommendations for appropriate chemical fertilizer use. Notably, our comparative analysis reveals that SVM achieves an accuracy of 88%, while ResNet surpasses with an accuracy of 94%. Consequently, the ResNet algorithm is selected for real-time implementation. |
Full text from SpringerLink