Fabric defect inspection is the pivotal part in the production of textile products. Since manual inspection is tedious and erroneous, automated fabric inspection has been topic of research for past years. Automation of fabric inspection involves two major aspects: defect detection and defect classification. We focused on classifying defects based on geometric features of defects. The features are obtained by applying statistical technique on an image dataset. Classification of defects is accomplished using simple Bayesian classifier, which delivers a pleasing accuracy.