การรู้จำลวดลายผ้าทอตีนจกแม่แจ่มโดยใช้โครงข่ายประสาทเทียมสังวัตนาการ

Abstract

Teen Jok woven fabric is a local handicraft inherited from the Tai Yuan, Tai Phuan, Tai Khrang and Tai Lue peoples, with unique patterns, colors and patterns that show the cultural prosperity of the Tai people. Mae Chaem Teen Jok Woven Cloth is one of the cultural handicrafts of the Tai Yuan people in Mae Chaem District, Chiang Mai Province that have been handed down for generations. This study was to compare the classification of 16 Mae Jam woven fabric patterns from 3 convolutional neural network (CNN) models, Which is EfficientNetV2B3, ResNet50 and MobileNetvV2. With Teen Jok pattern of Mae Chaem, Si Satchanalai, Nan, Laplae, and Muang Long in the amount of 800 images and divided into 20 classes, then enter the process of increasing the amount of images. and 6,000 new images were obtained. The image data was divided into 2 datasets, the training data and the testing dataset with a ratio of 80:20. The results of the classification in the testing dataset showed that EfficientNetV2B3 model has excellent performance in classification of Jok weaving pattern with accuracy of 80.94%, followed by ResNet50 and MobileNetV2 with accuracy of 73.54% and 37.71% respectively. This experiment proved that EfficientNetV2B3 model give an excellent model and has reliable results for analyzing images of the Mae Chaem Teen Jok Teen Jok woven fabric.