A user first uploads a picture of their face with a "neutral" expression. Then they use their mouse to mark the ends of their eyebrows, the corners of their mouth and the edges of their eyes and lips.What a waste of effort. Are chat programs so strapped for server space that users can't just upload one picture for each expression?
The software uses these points to morph the face to express different emotions: happiness, sadness, fear, anger, surprise, and disgust. A user can select an emotion and one of three intensity levels when using the system....
The software uses a facial image database to determine the correct spatial relationship between different facial features for different emotions. It distorts the points and areas marked by the user to shift them for a desired emotion.
Sep 30, 2006
ridiculous replacement for emoticons
Everyone knows that words on a screen lack context, which sometimes leads to horrific misunderstandings. Using pictures to replace :-{ (mustachioed regret) or ~:-) (toddler smile) seems like a good idea, until you realize how it's going to work:
Maybe the strategy is to discourage people from trying to game the system--uploading a sad picture when the Webbot prompts for happy, or uploading a different celebrity face or dog or piece of anatomy. Probably robots can tell what's a face, find a mouth and brows and make the crude adjustments that humans take for happy and sad (see Paul Ekman's work and software). Catching the gamers with user-supplied photos for all moods might require a lot more human vetting and labor cost. Sounds like a cool idea to me.
ReplyDeletePerhaps a compromise, then. Let users send multiple photos, but let the face recognition boot the ones that it can't "see." It's still better than the ridiculous faces the computer supplies.
ReplyDelete