Very interesting project and a nice poster. I am curious as to how your eye gaze detection would work if there are multiple persons present in the room.
Thank you for your comment! Due to the code limitation, gaze detection only supports searching for one face in the image captured. Hence we are planning to implement a possible solution to integrate sound source localization and face-tracking techniques into our system to locate where the user is.
The external servo together with the camera platform would then be mechanically turned to the user’s direction, leading the camera to spot the correct user. We decided to use this to solve the problems of where multiple users are in the same image. Despite the effectiveness of this methodology when the users are dispersed, it might still be inaccurate when the correct user is very close to other faces and so far we could not think of a way to solve so. We would appreciate it if you could give us any potential solutions or suggestions!
CHAN, Lok Him
January 19, 2022 2:07 pm
Greetings! We are from group SB04a-21, and we are pleased to present our midterm project poster to all of you! Unfortunately we are having some display problem for the poster, so please check this out for the rectified version. Also please feel free to ask us any question and we would be willing to answer!
Hi. Very nice work. I can see how the device could be of practical use.
Thank you for your appreciation.
Very interesting project and a nice poster. I am curious as to how your eye gaze detection would work if there are multiple persons present in the room.
Thank you for your comment! Due to the code limitation, gaze detection only supports searching for one face in the image captured. Hence we are planning to implement a possible solution to integrate sound source localization and face-tracking techniques into our system to locate where the user is.
The external servo together with the camera platform would then be mechanically turned to the user’s direction, leading the camera to spot the correct user. We decided to use this to solve the problems of where multiple users are in the same image. Despite the effectiveness of this methodology when the users are dispersed, it might still be inaccurate when the correct user is very close to other faces and so far we could not think of a way to solve so. We would appreciate it if you could give us any potential solutions or suggestions!
Greetings! We are from group SB04a-21, and we are pleased to present our midterm project poster to all of you! Unfortunately we are having some display problem for the poster, so please check this out for the rectified version. Also please feel free to ask us any question and we would be willing to answer!