HoikuCam: Voice Recognizing Camera for In Situ Recording of Child Development

We propose the HoikuCam, a camera with voice recognition, targeted at the usage in nursery schools. It starts recording the voice when the user turns on the camera, expecting the name of child is spoken. The recognized name is used to query a database to remind the expected forthcoming behaviors of the child, which is related to the current focus of the care. The voice recognition is also triggered when a photo is taken, which allows text annotationof the photo. The image is automatically classified by the child's name that appears in the recognized text. Our system significantly reduces the burden of caregivers when they write summaries of the care.