Human sensing is the science of detecting human presence, count, location, posture, movement, identity, and even behaviour from sensory data.
The need to sense people becomes ever more pressing, as the information is increasingly used to make decisions and provide services in diverse areas.
However, due to technological constraints, human sensing have not been applied yet in the field of television, but there is no doubt that once implemented – it would mean a real revolution, no less.
Pebbles Interfaces managed to overcome these traditional technological challenges, and developed a new technology that finally allows a new era of human machine interaction for mass market distribution.
This paper presents an overview of some of the potential new innovative capabilities which televisions would have, using Pebbles’ unique human sensing platform, and how it’s expected to transform the industry.
In the last decade we witness a proliferation of devices with advanced sensing capabilities that measure motion, orientation, various environmental conditions, and most recently – human sensing.
Human sensing is the extraction of information about the people present in an environment.
The data coming from the sensor provide indication about people presence, count, location, tracking, identity, and behaviour.
Using this data questions such as ‘How many people are in the room?’, ‘What is their position?’, ‘What is each person doing?’, ‘Are they comfortable?’ and so on can be answered.
Such human sensory information is used in many types of big data applications in various industries for better decisions making and to provide advanced services.
The applications range from simple implementation like open a door as people pass or lock a computer when the user goes away, to a more sophisticated usage such as biometrics-based user’s authentication.
Innovative human sensing has been adopted already in areas like security, medical, robotics and artificial intelligence, to name a few.
However, at this point of time, due to technological constraints, the usage is limited mainly to research and enterprises, rather than home or personal usage.
Therefore, human sensing have not been applied yet in the field of media (television, digital signage, etc.) despite the tremendous value it is likely to bring to this market.
Once human sensing capabilities could be introduced into the media market it would revolutionize this market and transform it from HW-based arena, to internet-dominate industry.
Targeted advertising, personal content delivbery systems and interactive engagement will be possible, offering an entirely new viewing experience.
In this paper we review the traditional approaches for human sensing and explain their limitations.
Then, we introduce Pebbles Interfaces unique technology, which enables for the first time, to break through existing barriers and allow a new era of human-machine interaction.
Afterwards, we provide a glimpse to the new world of opportunities that will be possible once Pebbles’ technology is adopted, and how it will transform the media market, and TV in particular.
A discussion of possible directions the market could evolve into is given subsequently.
TRADITIONAL HUMAN SENSING
When considering human sensing for the media market, there is a number of key parameters that should be taken into account.
These parameters include robust real-time high-resolution sensing capabilities within a certain range and field of view, which allow intuitive user interaction.
Low cost and small dimensions are also important. Therefore, only optical technologies are relevant to this discussion. Other technologies, which address just part of the human sensing requirements are not discussed in this paper.
Traditional optical human sensing optical based solutions include 2D RGB images / video and general depth mapping.
2D RGB Images / video
Compared to other sensors, cameras are affordable, and a great number of devices have high-quality camera by default.
Thus, camera-based human sensing usually requires a software-only solution and this field of computer-vision is highly developed.
However, person-detection or identification of person-related activities is a great challenge using 2D images.
There are several reasons for this challenge. The first is quite obvious – The world is three-dimensional and not two-dimensional, which means that understanding of 3D structures and z-axis movements from regular 2D image is not a straightforward task.
It may be doable but will require massive computing resources. In addition, there are many other fundamental challenges of 2D RGB images: it is difficult to distinguish between objects with a similar tone, movement may lead to smearing which impede the processing, and finally, lack of lighting causes data loss.