Abstract: Classification of particular group of agricultural images into semantically meaningful categories is a challenging task. Recently color coherence vector has become popular for image mining. This study makes use of multicolor coherence feature with multiple guide images (MGMCF) for classification of agricultural images like coconut and palm trees. The classification results using neural network is promising. Hence, image mining/image retrieval tasks can be done at good precision/recall by using MGMCF features.