Deep Machine Learning System for Athletes
Hong Kong Univisual Intelligent Technology Limited (HKUIT) is founded by scholars and graduates of the University of Hong Kong. It applies Artificial Intelligence (AI) and vision sensing technologies to the sport industry. Prof. Wilton Fok is the co-founder of HKUIT. Prof. Fok is also the Director of the Sport Artificial Intelligence Laboratory of the University of Hong Kong, also the Assistant Dean and the Chief Lecturer of the Faculty of Engineering at the University. His team has created a Drowning Detection System for indoor swimming pools to monitor swimmers who may be in danger of drowning. The system can be adapted to posture correction for athletes to improve their sport performance.
Drowning Detection System protects swimmers
HKUIT is applying for two patents for their swimming safety and swimming posture analysis system. Those patents are related to their Drowning Detection System.
This Drowning Detection System has incorporated AI technology with cameras to identify human-like objects in water who have been immobile for a certain duration. Cameras are installed upon the ceiling, by the poolside and also under the water. Images of the postures of the swimmer will be captured by those cameras in different angles and transmitted to the graphics processing units (GPU) and the centralized console for data processing. Data such as arm-stroke, leg kicking, arm circulating frequency, arms and feet coordinating patterns, forward speed and amplitudes are collected for analysis in a computer at real time. When a human-like object becomes stale in the water, the Drowning Detection System is alerted and its audio and visual alarms will be triggered. Underwater image will display on the screen of the system. Lifeguards will be notified and he can use a tablet computer to identify where the potential drowning person is.
Applying Human Pose Estimation Technology in Sport
The Drowning Detection System is fitted with Human Pose Estimation Technology. It can therefore also be used for swimming coaching. Utilizing cameras, the system can identify 13 skeletal nodes of the human body and monitor those movements dynamically. Data of professional swimmers are stored into the system. Jumping and swimming posture of a swimmer will be captured by the cameras in the pool. The embedded AI technology will compare the data and differentiate the motions between the swimmer and the professional in real time. The coach and the swimmer can watch the video in slow motion on a computer and study the report generated by the system together. The system performs detailed analysis on training progress, provides a custom-made training routine and makes suggestions so the swimmer can achieve a higher level. Using AI technology to obtain objective data for sport training, the training can be more effective. Professional athletes can achieve greater performance. In the 2021 Olympic 100 meters freestyle final game, the Hong Kong female swimmer and the silver medal winner, Siobhan Haughey, competed fiercely with the Australian gold winner and came just 0.31 seconds behind. If she can improve a little bit with AI technological assistance, the outcome is the difference between the gold and silver medals.
Currently, the system requires more data to achieve the best result. Because of sunlight reflection under the water, the AI needs data to learn and distinguish shadows. In addition, more swimming posture data from different swimmers are required to build a better analysis program for the swimming training system. With thousands or more image data for machine learning analysis, the computer can learn what is good swimming skill and what is bad. These devices have already been installed in the swimming pools of the Hong Kong University and two secondary schools in Hong Kong. Hopefully more local schools will adopt the technology.
AI golf coaching system
HKUIT has also applied its Human Pose Estimation Technology system to golf and other sports. The posture of golf players determines their result directly. The right body swing is critical to the performance. Robocoach is an AI golf coaching system invented by HKUIT. It is a mobile application and will be launched in September 2021. Robocoach will capture the golf player movement via the camera on mobile phones. It can identify his/her 13 skeleton nodes and record the body motions. Image data will be transmitted to the deep machine learning system to analyze the body movements in order to assess the level of power, to grade on each swing and to provide feedback to players. For instance, too large arm swing motion or improper twist movement can be identified by the system. Following the instructions of Robocoach, players can improve their posture to produce powerful swings consistently.
The next phase of Robocoach is to enable multi-players participating in a golf game and host competitions in the virtual world. Players will have more fun playing with friends in different locations.
Empowering sport activities by AI
HKUIT has been partnering with a fitness company to provide yoga courses and exercise courses in the form of online videos through a mobile application. They are devoting their expertise and energy to develop more smart physical exercise equipment for people to workout safely with joys. As the world celebrates the success of the 2021 Olympic, we all look forward to more application of human posture technology development in various sport activities such as fencing, dancing, basketball, volleyball, etc.
Written by Ms. Florence Ip. Edited by: Dr. Kyle Wong
Please contact ACH to learn more:
ACH Worldwide Limited Website: www.ach-worldwide.com
ACH articles: http://www.ach-worldwide.com/blog.html
ACH Worldwide Limited Email: contact@ach-worldwide.com
ACH Worldwide Limited LinkedIn: ACH Worldwide Ltd
