TOP  >  Thesis/Dissertation  >  Detection of the timing to which it can't take one's eyes off work

Detection of the timing to which it can't take one's eyes off work

In the task that the teacher in the remote place teaches the student the work method through the image communication and voice distribution, the teacher's instruction should not disturb the student's work. Then, in the cooking experiment, we analyzed timing to which the student cannot take his/her eyes off work. Consequently, "the hand is arranged for the following work" operation was done at timing to which the student can't take his/her eyes off work. Because the student didn't completely understand the work space, the student had to con rm the position and the shape of the object by eyes for the following work. 72.14% of such timing that had been observed in the cooking image was "Reach it for the movement of the object" operation and "The hand is arranged to cut the ingredient with the kitchen knife" operation. We considered what technique was effective for detecting these two kinds of operation by applying variously existing techniques and comparing the detection results. As a result, it has been understood that it is important to extract the hand area accurately. In the hand area extraction, TexCut that dealt with the illumination variation obtained accuracy that was higher than grabcut widely used for the foreground area extraction.