Hand-Woven Fabric Motif Recognition Methods: A Systematic Literature Review

Knowledge of hand-woven motif recognition is only owned by the older generation, which has not been passed down to the younger generation. Moreover, computer technology can be used to support the recognition of traditional woven fabric motifs. The aim of this research is to conduct a literature review on hand-woven fabric motif recognition, with an emphasis on performance accuracy, techniques, and datasets utilized. This research included three distinct stages of systematic literature review (SLR): planning, execution, and reporting of findings. We found several datasets between 924,845 data and the number of classes between 3–25 class. Based on research result, we obtained several methods for hand-woven fabric motif recognition focused on performance examination. We recommended image pre-processing method, including Adaptive Filtering Denoising, Adaptive Wiener Filtering, Histogram Equalization, Gradient Pyramid (GP) Decomposition. Moreover, we suggested five feature extraction methods, including Radon Transform, Wavelet Transform, Locally Rotation Invariance Measure (LBP-ROR), Transform Invariant Low-Rank Textures (TILT) and Histogram of Oriented Gradients (HOG). For learning method, we recommend Fuzzy C-Mean (FCM), Convolutional neural network (CNN), Deep Neural Network (DNN), MobileNets, Inception-v3 and ResNet-50.