Face Recognition on iOS & Computer Vision: Demo | Sighthound.com
The iPad is trained by a 3-second face capture. In 30-45 seconds, the deep network extracts features and saves them into a gallery. As Joy enters and comes into view, the robot will recognize her. [ROBOT]: Hi, Joy. If the face is not recognized, a red box appears and a custom greeting is dictated. [ROBOT]: Hi, welcome to Sighthound. Please sign in. The software is a full-convolutional, deep neural network [ROBOT]: Hi, Joy. running locally on an iPad using only the CPU. Detection, tracking, and recognition processes are all running in real time on the device. The software even continues to track a person with partial occlusion.