基于知识图谱构建的面料图像多样化检索系统OACSTPCD

In order to improve the efficiency of fabric selection in the textile industry,satisfy diverse retrieval needs of users and solve the problems of single fabric retrieval result,long time consumption,low accuracy in fabric retrieval and so on,the text information was structurally represented through knowledge modeling,and a fabric image and text dataset were constructed.Visualization of fabric knowledge graphs was achieved by importing the graph database.Based on the knowledge graph,a fabric image diversified retrieval system was established.The retrieval fields provided by users were matched with fabric nodes in the knowledge graph by the system.The corresponding fabric entities and their first-order nearest neighbors were output.Diversified fabric image retrieval was achieved.50 text keywords were selected for retrieval experiment.The results showed that the average precision was 80.7%,the mAP value was 0.852,the average diversity value was 5.8 for the first 8 images.The average response time of retrieving was only 2.26 s.The effectiveness and feasibility of the system were validated.