From voting evaluation to gaze evaluation

The voting system introduced on the page of the collaborative design support system, but We found a serious problem while proceeding with the research. It's surprisingly difficult to get the number of votes.This system assumes that users will actively vote, so the design will not evolve unless the number of votes. To solve this problem, we proposed a system in which designs are displayed on digital signage in the city and people can vote on the spot, as shown in the figure below, but the process of stopping and voting (i.e., pushing the voting button) seems to be annoying, and it was difficult to get votes.


Landscape of digital signage voting system

Although these systems have achieved some results in the evaluation experiments, we think that further improvement is still necessary.

For improvement of the system, we are verifying the use of gaze information. When using digital signage, if each user just stops and looks at the design instead of voting, or also looks at the design while walking. We would like to build a system that applies the real environment.

The figure below is an overview of the system that creates character coordination based on the gaze information of multiple users.


Overview diagram of the Character Coordination System

In this system, the characters appear in various coordinates and move for about 10 seconds. During that time, multiple users can freely view the character, and the system evaluate the coordination based on the gaze information at that time. The design parts used for coordination are as follows.


Design parts for character coordination

The coordination evaluation of this system uses the winning paired comparison evaluation method introduced on the page of the men's watch design system, and the winner coordination is determined by the gaze information of multiple users.

We confirmed various effectiveness from the results of evaluation experiments that even if some of users in a group partially participates in a round evaluation for a series of round evaluations, the system can generate a satisfactory design to group of users.
Related our works
[Academic Papers] Hiroshi Takenouchi, Masataka Tokumaru, "Character Design Generation System Using Multiple Users’ Gaze Information", IEICE Transactions on Information and System, Vol.E104-D, No.9, pp.1459-1466, 2021-09.
[International Conference] Minatsu Fujisaki, Hiroshi Takenouchi, Masataka Tokumaru, "Developing Female Clothing Coordination Generation System Using Eye Tracking Information", Human-Computer Interaction. Interaction Technologies Volume 10903 of the series Information Systems and Applications, incl. Internet/Web, and HCI (the proceedings of HCI International 2018), pp.247-257, 2018-07 (Las Vegas, USA).
*Bold: Students in this research laboratory

▲pagetop