Making a Smart TV smart about us

Making a Smart TV smart about us

Making a Smart TV smart about us

A Smart TV should be able to understand what happens around them to provide more personalised show schedules, avoid irrelevant advertisement, block channels the current viewers should not watch, and so on. Such a Smart TV would also be able to provide better parental control features, children activity monitoring, and adaptive sound and lighting.

A Smart TV must be able to understand if one or more persons are present, what they do and the type of interaction with the TV, which can be classified into focused (TV is the primary activity), monitoring(TV is secondary activity) or idling  (TV is simply background).

We need to know what people do

To get further insight on what is truly happening in front of the television, we need to be able to monitor the entire area in front of it. This can be done to a very limited extent with technologies such as cameras. However, these are considered intruding personal privacy as demonstrated by the recent backslash against smart TV equipped with them. Using information coming from personal devices nearby, like smartphones or smart watches, has also shown to be ineffective. Unfortunately, when at home they are often not in the proximity of the viewer and activity on them cannot be captured.

Simple temperature sensors can be enough

At Xetal we have developed a technology able to detect presence and position people in real time. The detection accuracy is high enough for understanding person activities such as walking, standing, sit on a sofa and so on. The technology uses only simple temperature sensors that can be embedded in existing devices such a TV set, a decoder or a TV Sound Bar. The technology is non-intrusive and allows to extract data without privacy concerns.

As an illustration of its use, a recent study was conducted using the new Xetal Development kit on a family of four (two adults and two kids). Activity was monitored when the TV was active. The figure above shows one of the several data set obtained. It illustrates the probability of viewers to be present when the TV is active. We can easily recognize two position where two viewers are most likely sitting and two most likely laying down. By combining such positional data with data on the displayed content, we have been able to determine that the sitting viewers are the children and the laying ones are the adults. Furthermore, the study has shown how children tend to watch TV more attentively that the adults, who prefer to lay down and use portable devices like tablets. Trajectory data has also been used to evaluate the TV set was active but nobody was watching, something happening mostly with adults.

This is just an example of how the information provided by a Xetal system could be used to determine not only presence and number of people, but also their interaction level without invading their privacy.

About the Author