Efficient detection of defects in fabric plays a dominant role in automating quality control. It significantly enhances efficiency as well as accuracy of quality assurance in textile manufacturing. Skilled humans who physically examine defective designs in numerous sectors use traditional defect detection techniques. The main drawbacks of manual methods include deficiency of concentration, inaccuracy, weariness and increased time consumption. Computer Vision (CV) and Digital Image Processing (DIP) techniques may be efficient in overcoming these shortcomings. The proposed approach involves data acquisition to record images which are subsequently pre-processed using histogram equalization to improve the defective area. A distinctive Neural Network (NN) is used for defect detection. This method improves precision and efficiency in identifying defects. The proposed method is built using Adaptive NNs (ANNs) along with Radial Basis Functions (RBFs) and is aimed at revolutionary identification of fabric flaws. The networks are built using attributes derived from images segmented using Expectation Maximization (EM) depending on target’s region attributes. As a result, the type of defects is classified based on the extracted features, thus yielding increased detection accuracy.