In this day and age spatial image classification has become a challenging task. Many systems are developed based on the different approaches of retrieval but the need of image based retrieval system is of dire need. It takes into account an image as the input query and the images are retrieved based on content of the image which is a complicated task. For this reason, CBIR system is used as an approach for retrieving semantically-pertinent images from an image database based on automatically-derived image features. The Objective of this Paper aims at retrieving images from spatial technology by taking some soil locations and classifying them using Adaptive resonance theory (ART), a Neural Network model for more proficient and successful results. Hence, with help of ART forthcoming problems or disasters can be easily studied and predicted which helps not only in rescuing mankind but also mother earth in priory.