Toprak türü sınıflandırması için görüntü tabanlı makine öğrenimi ve ürün stratejisine yönelik uygulamalar

Soil classification is a vital task in various fields such as agriculture, construction, and environmental studies. The present thesis harnesses the power of Artificial Intelligence (AI) to develop a robust and efficient soil classification system. Utilizing a curated dataset of 592 images representing five different soil types, the study explores different deep learning (DL) frameworks, including Convolutional Neural Networks (CNNs), InceptionV3, and MobileNetV2, to process and classify these soil images. The methodology involves a pre-processing phase, ensuring uniformity across the dataset, followed by the application of the selected models. Among the models tested, MobileNetV2 stood out, achieving an unparalleled accuracy rate of 95%, while CNN and InceptionV3 each reached 85% and 93% accuracy, respectively. The comprehensive evaluation included the use of confusion matrices and other performance metrics, validating the exceptional proficiency of the MobileNetV2 model. This work not only advances the field of soil classification but also illustrates the broader applicability of AI and DL techniques in environmental analysis and beyond.