Using image retrieval and computer vision techniques, a pottery shape and school classification system for labeling unknown pottery and pottery fragments was developed to assist archaeologists in identifying and recording objects quickly and accurately. The system can identify twenty pottery shapes and four pottery schools with shape and color-based image retrieval techniques. The system analyzes and compares extracted features to determine the five closest database images, and then presents the results to the user. This is the first pottery study to combine two different techniques---shape and color-based image retrieval---in identifying multiple characteristics of an unknown pottery image or pottery fragment. Experimental verification of system performance utilized two databases---a training database of two hundred digital images of twenty different pottery shapes and four different schools, and a testing database of four hundred images of equally distributed pottery shapes and schools. Four major areas were explored and the following first-choice accuracies obtained on the four hundred test images: 100% pottery school identification, 97.50% pottery shape identification, 96.25% accuracy of template-matching related to shape, and 69.5% accuracy of template-matching related to decoration. The first two of these accuracies were based on images of intact pots while the other two were on images of portions of pots.
