Database Reference
In-Depth Information
Fig. 4.9
Different gestures for specifying user intent in
TapTell
:(
a
) “tap”—selection of image
segments, (
b
) “line”—rectangular box, and (
c
) “O”—circle or lasso
indicated by user's gesture will be connected as the region-of-interest (ROI). The
ROI will be further used as the visual query, as shown in Fig.
4.9
a.
Line.
A user can draw straight “lines” to form a rectangular bounding box. The
region in the box will be used as the visual query, as shown in Fig.
4.9
b.
O(
circle
).
A user can naturally outline an object of irregular shape. The “O”
gesture can be also called the
circle
or
lasso
. Note that an “O” is not limited to a
circle, but any arbitrary shape, as shown in Fig.
4.9
c.
A user study is performed following the principles of focus group in the field
of human-computer interaction [
118
]. In this study, ten participants were invited.
After being introduced to the basic functions of
TapTell
and getting familiar with the
system, they were asked to perform several tasks using different gestures in 30 min.
From this study, it is found that seven out of ten subjects thought that “O” is more
natural than the other two gestures, and eight subjects were satisfied with the “O”
interaction. Their comments on “tapping” and “line” are: (1) tapping is sometimes
too sensitive and image segmentation is not always satisfying, and (2) the “line” is
not convenient for selecting an arbitrary object.
Equipped with the “O” gesture and the user interaction platform, mobile search
and recognition can be achieved effectively using the context-embedded visual
approach. The next step of
TapTell
is to recommend social activities based on
associated metadata and text-based search.
Search WWH ::
Custom Search