Sighthound Cloud brings accuracy to emotion and gender detection
Palo Alto, Calif.-based Sighthound has released an emotion recognizer on its Sighthound Cloud API for developers.
The recognizer has a Detection API for person, face, gender, and facial landmark detections, and a Recognition API that developers can use for face and vehicle recognition applications.
Its Cloud Recognition API makes it easy to add face recognition to applications. Currently, the API only supports Objects with a type of "person," but future updates will include additional types.
Developers can upload images of the people to be recognized, adding those people to a group, training the group, and finally confirming that the recognition is working by testing against some new photos containing the people.
The Sighthound Cloud REST APIs return the location of any people and faces found in an image. Faces can be analyzed for gender and identified against a gallery of developer-provided images.
A landmark detector will find the various facial features in the detected faces, including eyes, nose, and mouth, by fitting 68 landmark points to those features.
The emotion recognizer analyzes an image or video stream to find all faces and determines for each face the emotion along with an associated confidence level. Its emotion recognizer has been trained to recognize seven different types of emotions.
Sighthound says that it has the world's "most accurate age detector, gender recognizer, and emotion classifier" It says it provides "significantly more accurate results than the Microsoft Cognitive Services API."
Its "Real Age estimation" absolute error is 5.76 against Microsoft Cognitive services API (7.62), which is 1.86 years more accurate. Its "Emotion Recognition accuracy" is 76.1 percent compared to Microsoft's 61.3 percent.
Its gender recognition is accurate 91 percent of the time compared to Microsoft Cognitive Services accuracy, which is accurate 90.86 percent of the time.
Sighthound has also released a demonstration version of the emotion recognizer tool on its website so that users can upload images and test results. There are also tutorials, sample code, and documentation to help integrate its APIs into developer applications.
Developers can use the free Sighthound Cloud API for up to 5,000 API calls. Paid accounts (Basic, PRO, and Enterprise) have access to production servers for commercial use and begin at $49 per month.
VIDEO: Flippy the robot uses AI to flip burgers at fast food joints