论文部分内容阅读
通过实验采集用户的图像标注结果,对3种图像语义标注模式——基于标签打分的图像标注模式、单标签下基于图像比较的标注模式以及多标签下基于图像比较的标注模式的标注效果进行对比研究。研究发现:1基于标签打分的图像标注模式和单标签下基于图像比较的标注模式能够帮助用户对图像各标签的语义强度进行有效标注;2多标签下基于图像比较的标注模式可以帮助用户对图像各标签语义强度的比例关系进行有效标注;3标注界面中是否同时显示图像的所有标签,可能影响到用户对图像在各标签上语义强度比例关系的判断。
Through the experiment, we collected the user’s image annotation results and compared the three image semantic annotation modes - the image annotation mode based on label scoring, the annotation mode based on image comparison under single label, and the annotation effect of annotation mode based on image comparison under multi-label the study. The research found that: 1 tagging-based image tagging mode and single-tagging based on image comparison of the tagging mode can help users semantic strength of the image tags effectively tagging; 2 tagging based on image comparison tagging mode can help users to image Label the semantic strength of the ratio of the effective annotation; 3 marked interface in the display of all the labels at the same time, may affect the image of the user in the label on the semantic strength of the proportion of judgments.