<![CDATA[A digital image-based leaf recognition system is one of the modern solutions in the fields of botany and agriculture to identify plants automatically. This study developed a leaf recognition system using image processing methods and Convolutional Neural Network (CNN) algorithms. CNN was chosen because of its ability to independently extract features through convolution layers, thus capturing important visual patterns such as shape, edges, textures, and leaf veins without requiring manual feature engineering processes. The research dataset consists of a collection of leaf images from several types of plants obtained through direct photo-taking and public dataset sources. Each image goes through a pre-processing stage, including cropping, resizing, image quality enhancement, and pixel normalization to ensure data consistency before entering the training stage. The CNN model is designed with several convolutional layers, pooling, activation functions, and fully connected layers to produce optimal classification performance. Model training is carried out by dividing training and testing data, as well as augmentation techniques to increase image variation. System performance is evaluated using accuracy, precision, recall, and confusion matrix. The test results show that the CNN model is able to recognize leaf types with a high level of accuracy and is stable under various test conditions, including variations in lighting and shooting angles. Overall, this study proves that CNN is an effective and reliable approach in building an automatic leaf recognition system. This system has the potential to be applied in the fields of precision agriculture, mobile application-based plant identification, and botanical research that require speed and accuracy in plant classification.]]>
