Basis Pengetahuan Nilai-nilai Kain Tenun Sumba dengan Model Seci dan Convolutional Neural Network

Abstrak

This study aims to make tacit knowledge explicit about the philosophical values contained in each pattern of Sumba woven fabrics. Loli woven, white textile, Lambaleko fabric from Lamboya, Panggiling textile from Wanokaka and Lamboya, and modification Panggiling from Wanokaka are types of cloth in West Sumba Regency, especially in four sub-districts. This study detects the type of fabric from each existing motif. This study uses the Seci model combined with CNN and is presented on the website. From this research, the tacit knowledge of Sumba woven fabrics can be made explicit. The results from CNN show that of the total epochs built, the accuracy value at the 10th epoch is 80%, the accuracy validation is 100%, the accuracy error value is 0.2638% loss, and 0.0477% validation loss. The high accuracy validation results, low accuracy errors, and low validation loss are caused by the image data used as test data, which is a type of motif with distinctive characteristics, in contrast to image data of plants and other living things, which have various types. Website development as a user interface has been going well, as evidenced by the success of predicting tacit knowledge for each image used as test data.