CLASSIFICATION OF DURIAN LEAF IMAGES USING CNN (CONVOLUTIONAL NEURAL NETWORK) ALGORITHM

This research investigates the classification of durian leaf images using Convolutional Neural Network (CNN) algorithms, emphasizing the importance of accurately identifying different durian varieties based on their leaves due to their distinct characteristics such as taste, texture, and aroma. The study aims to evaluate the performance of three CNN architectures AlexNet, InceptionNetV3, and MobileNet in classifying images of durian leaves from five classes: Bawor, Duri Hitam, Malica, Montong, and Musang King. A dataset comprising 1604 images for training, 201 for validation, and 201 for testing was collected. Data preprocessing involved steps like data augmentation, pixel normalization, and resizing images to 150x150 pixels. The CNN models were trained using TensorFlow and PyTorch frameworks, and their performance was evaluated using a Confusion Matrix to assess accuracy, precision, sensitivity, specificity, and F-score. The results revealed that InceptionNetV3 and AlexNet achieved near-perfect classification accuracy, while MobileNet, despite showing high accuracy, exhibited some misclassifications, indicating the need for further refinement. In conclusion, InceptionNetV3 and AlexNet proved to be highly efficient and accurate for classifying durian leaf images, making them suitable for practical applications. MobileNet, though performing well, requires additional tuning to achieve similar accuracy and reliability. This study highlights the significance of selecting appropriate CNN architectures and thorough preprocessing to optimize model performance in image classification tasks.