The productivity of the Indian agriculture is largely based on the effective use of land and crop selection. Effective identification and selection of suitable land and crops can improve productivity and contribute to sustainable agriculture. The current study guides the farmers in several aspects that include land categorization by satellite images and analyzing soil nutrients The proposed system is adaptable in various Indian agricultural zones and enable the effective decision-making. The approach employs different pre-trained models like VGG16, VGG19 and ResNet50 to classify a set of satellite and aerial images. The results obtained state that the VGG16 achieves 92.7% accuracy towards the classification of land type followed by 91.09%, 94.10% using VGG-19 and ResNet50 respectively. Further analysis is done towards grooved fields with soil nutrients (N, P, K), temperature, humidity, pH, and rain levels. The current data is trained with different machine learning algorithms like Random Forest, Logistic Regression, and Decision Trees to recommend the suitable. The implementation of algorithms produces accuracy of 99.55%, 97.27%, and 98.64% respectively. Instead, strong preprocessing including feature scaling and label coding is employed to make performance of models more effective. A light web-based user-friendly application has also been developed that provides the users with real-time crop recommendations based on land images and other environmental features. The integrated system enables data-driven agriculture and guarantees an improved efficiency of yield without resource wastage.