<![CDATA[Weather is a determinant of farmers' planting calendar. Farmers usually start planting rice in the rainy season because rice requires sufficient water to produce optimal harvests. The weather is almost unpredictable in certain months, so farmers now look at cloud conditions to predict the season. Seasonal predictions based on cloud imagery can be assisted using Artificial Intelligence methods. Previous research used deep learning via transfer learning, but the results were not optimal. This research dataset is sourced from Kaggle and consists of five classes, namely cloudy, foggy, rainy, shine, and sunrise with a total data of 1500 images. This research proposes that a hybrid deep features and machine learning approach be used to increase the accuracy of the results. The MobileNet deep learning method is used at the feature extraction stage, then for classification using the Support Vector Machine (SVM) method. Experimental results with the Radial Basis Function (RBF) kernel on SVM produced an accuracy of 0.9500 for training data. The evaluation results using testing data produced an accuracy of 0.9667. This result also saw an increase of 4.2% in training data compared to previous research. Through these results, MobileNet-SVM is proven to be able to improve classification accuracy when using a small dataset with 1500 images.]]>
Copyrights © 2026
