face
Researcher verifies eye-guiding from Rembrandt
Researcher Steve DiPaola from the University of British Columbia has scientifically verified the "eye guiding" technique on Renaissance art. He attributed the origins to Rembrandt. The technique has been used to guide a viewer around the portrait and creating a narrative as a result. DiPaola used an eye tracking system to track viewers' gaze around portraits, consisting of original Rembrandts and another from computer-generated versions with different focus on specific areas of the face.
Google weighing whether to offer face recognition
According to the Financial Times, Google is deliberating whether to offer face recognition capability, and if so, how to address privacy issues. Google has received criticism about privacy issues in other areas, for example, related to its roving camera vans that support its StreetView service.
Grad student foils face detection
An article in The Register reports that New York University grad student Adam Harvey has designed makeup patterns and disguises to foil face detection software. Mr. Harvey studied the patterns learned by the Haar Cascade method used in OpenCV's face detector, and created makeup patterns that cause faces to be missed by that detector.
Toshiba captures dynamic 3D face models
Submitted by Boaz Super
Toshiba's Cambridge Research Laboratory (CRL) has developed a computer vision method for capturing dynamic 3D face models. Red, green, and blue lights simultaneously illuminate the face; a single camera measures the mix of colors reflected by each surface point to compute the orientation of that part of the face. The technology could be used to generate high-quality avatars.
Vision based lie detector identifies physical signs of guilt
Submitted by Sek Chai
Researchers from Aberystwyth University and the University of Bradford have developed a lie detector, called Real-Time Dynamic Passive Profiling, to identify threats from smugglers and terrorists at border control points. The system uses a thermal-imaging scanner to detect signs of guilt by modeling "facial expressions, eye movements, and pupil changes in both visual and thermal domains".
UMass Amherst reports on emotion-sensitive, computer-based tutor
University of Massachusetts Amherst researchers have reported results from their computer-based tutors in a press release. Their system uses a camera to detect facial expressions to determine users’ moods, at about 70 to 80 percent accuracy. The researchers found that girls in fifth grade "appreciate the emotional support that matches their mood that is given by the computer software, but it seems less important to boys."
Computer vision smiley generator: just try not to laugh
Submitted by Boaz Super
Theo Watson at F.A.T. (Free Art and Technology) has demo'd Auto Smiley, which inserts smiley emoticons into email or chat whenever it detects the user smiling. This is a fun project, but try not to laugh, unless you want lots of smileys in your email.
Auto Smiley was built as a speed project in one hour using openFrameworks and the Machine Perception Toolkit.
Motion tracking and facial animation combined in partnership of two companies
Xsens Technologies, which sells an inertial motion capture suit, and Image Metrics, which provides facial animation services based on computer vision methods, have partnered to provide a complete body and face animation solution. The combined solution allows actors full freedom of motion. The technology is already in use by a special effects studio, Double Negative. More information is available in a press release.
TAT Recognizr identifies faces in mobile phone augmented reality app
Swedish company The Amazing Tribe (TAT) has developed an Android smart phone app, called Recognizr, that identifies a person when the phone's camera is pointed at them. Using face recognition software developed by another Swedish company, Polar Rose, the augmented reality app displays information about the person superimposed near the image of that person. Only users who have opted into the system and provided a face image and profile are identified by the system.
New study provides support for modularity theory in human vision
A joint study by researchers at Beijing Normal University and MIT has found that the ability to recognize faces has a genetic component, and that it is uncorrelated with IQ test scores intended to measure 'general intelligence.' According to Professor Jia Liu, the results provide the first genetic evidence for the modularity hypothesis of mind. This work has potential implications for the development of computer vision systems. More information is available from a Medical News Today article.
