Image Retrieval of Indonesian Batik Clothing Based on Convolutional Neural Network

Abstract

Indonesian Batik is best-known for unique and distinct pattern. Searching Indonesian Batik clothing images is a challenging problem due to its wide pattern variations. In this paper, proposed image retrieval model of Indonesian Batik clothing image searching based on Convolutional Neural Network (CNN). Autoencoder proposed as CNN model that trained to reconstructed original input batik clothing image. So, the visual features can be extracted from CNN Autoencoder. Based on the experimental results, the proposed method can reach 90.8% in retrieval accuracy, 58.8% in mean average precision, and 61.9% in average recall.