Recognizing Motions and Improper Postures

ETRI has developed a motion-recognizing technology by using 'machine-learning' technology that is gaining attention in recent years.
This technology recognizes a user's workout motions in real time and automatically diagnoses any improper postures.
The machine-learning technology is part of artificial intelligence with which computers independently learn and
find patterns from data and serves as a core technology behind a range of services announced at Google I/O 2015, held in June.
The expression 'learning' is used because computers voluntarily learn and find patterns from a vast volume of data.
The development of learning techniques in recent years has resulted in research outcomes showing
how computers have exceeded human intelligence in a range of areas.

Smart Golf Coaching/Training

ETRI has applied this technology to golf.
If a user tries a swing movement, the computer will automatically recognize the beginning
and ending of the swing and first show a front image, followed soon after by side images.
It then compares those images with the front and side images of a professional golfer
and analyzes the movement of the user's head, shoulders, and feet.
With only one camera in the front, the computer can conduct a detailed analysis of even the side postures
and accurately identify the angle of the body during the swing movement.
This technology analyses the movement of a user's joints by means of
a depth-sensing camera that acquires 3D spatial data.
The computer uses a progressively darker red color to demonstrate a range of error data,
such as the head moving during a swing, improper angle of the arms, or wrong shift of the center of gravity,
as a means to help users immediately understand their wrong postures.

Wider Use Expected in Sports and Medical Sectors

The team of researchers is planning to focus on developing a high-performing
motion-recognizing camera that exceeds the capacity of existing products.
Following technology transfer to Maum Golf, ETRI has launched commercial service of
the smart golf coaching/training program, the first of its kind in the golf industry
by combining its technology with the golf lesson program offered by Maum Golf School.
ETRI's motion-recognizing technology based on machine-learning is expected to be widely
used not only for a variety of sports, including golf, virtual fitness and baseball, but also for medical services,
most notably rehabilitation treatment, which requires precise analysis of motions and postures.
Dr. Ji-Young Park of ETRI's Interactive 3D Image Research Laboratory stated,
“ETRI will provide full support in helping this technology transfer lead to successful commercialization of this system.”
The machine-learning technology was developed through joint research, as part of the state-of-the-art
convergence content technology development project supported by the Ministry of Science, ICT and Future Planning.