Specifying the relationships between objects, gaze, and descriptions for scene understanding
Specifying the relationships between objects, gaze, and descriptions for scene understanding
复制标题
指定物体、凝视和描述之间的关系以进行场景理解
DOI：
--
复制
发表时间：
2013
期刊：
影响因子：
--
通讯作者：
G. Zelinsky
中科院分区：
文献类型：
--
作者：
Kiwon Yun;Yifan Peng;Hossein Adeli;Tamara L. Berg;D. Samaras;G. Zelinsky
目标·进行结合行为和计算机视觉的实验，以更好地理解之间的关系：识别图像中检测到的对象，识别人们在观看图像时的眼球运动，识别和他们在被要求描述图像时产生的词语。贡献·理解人类如何看待和解释视觉图像。·通过将凝视线索与当前视觉识别系统的输出相结合，演示凝视启用检测和注释的原型应用程序
Goal • Conduct combined behavioral and computer vision experiments to better understand the relationships between: Ø the objects that are detected in an image, Ø the eye movements that people make while viewing that image, Ø and the words that they produce when asked to describe it. Contribution • Comprehension of how humans view and interpret visual imagery. • Demonstrate prototype applications for gaze-enabled detection and annotation by integrating gaze cues with the outputs of current visual recognition systems