The capital battle of 5G is fiercely fought, with major telecommunications companies establishing 5G alliances one after another. In addition to bringing in equipment manufacturers like Nokia and smartphone makers like Sony and Samsung, Taiwan's leading telecom company, Taiwan Mobile, has this time enlisted a group of featherweight startups with an average staff of fewer than 20 people.
Taiwan Mobile's General Manager, Lin Chih-chen, said, 'Just the business opportunities in 5G alone are worth at least trillions of New Taiwan Dollars. However, with so many applications across various vertical industries, it's challenging for telecom companies to go it alone. What is the background of these newcomers, and how did they come up with these interesting new applications?'
Now, when you walk into four branches of the Shanghai Commercial and Savings Bank to withdraw money, you may not be aware that cameras are detecting whether you are nervous or have an abnormal heart rate. If so, the system will alert bank employees to come out and inquire if the customer might be a victim of fraudulent money transfers.
This is the key technology that led FaceHeart Inc. to be selected into Taiwan Mobile's 5G alliance. This small software startup with just 20 people can measure heartbeats by observing changes in the color of facial micro blood vessels, and they even secured investment from MediaTek this year.
Huang Chung-hsien, the General Manager of FaceHeart Inc. Intelligence, who previously worked at MediaTek for 18 years, mentioned that common physiological monitoring technologies nowadays are "contact-based," such as using a smartwatch to measure heart rate. However, FaceHeart Inc. conducts "non-contact" physiological monitoring through facial images captured by cameras. The advantage is that it eliminates the need for expensive and sophisticated medical equipment. 'Within 6 seconds, we can tell you what your heartbeat looks like,' said Huang Chung-hsien.
There are numerous teams abroad conducting research on heart rate detection through imaging, such as the U.S. startup team Smart Beat, which has introduced a camera and app for detecting a baby's respiration and heartbeat to prevent sudden infant death syndrome. In the intense competition, accuracy is the key factor. Currently, under conditions of remaining still, FaceHeart Inc. measures the heart rate with an error margin of plus or minus 3 beats per minute and is still improving.
This artificial intelligence technology for heart rate detection using image recognition was initially developed by FaceHeart Inc. founder and special professor of the Department of Electrical Engineering at National Chiao Tung University, Wu Bing-fei. About 5 to 6 years ago, he and his students began their journey from experiments in recognizing fatigued drivers through image recognition. To collect facial images and physiological data of drivers, they embarked on a road trip from Hsinchu at 3:00 PM, drove all the way to Nantou, and returned before 9:00 PM. The students had cameras installed in their vehicles and wore heart rate monitors on their fingers, recording the changes in facial conditions and heart rate during the 6-hour trip under different lighting conditions of day and night.
After several years of research, he accumulated data from several hundred people and nearly ten thousand records to develop this algorithm model.
Interestingly, this technology can be applied not only to remote medical records but also to various scenarios, such as detecting driver fatigue on a bus. As the era of 5G begins, technology can simultaneously transmit larger quantities of high-definition video footage and provide faster feedback on recognition results. In the future, for remote and real-time medical consultations, a camera pointed at the patient will allow doctors to instantly know the patient's heart rate and blood pressure. This will accelerate the implementation of future smart healthcare!
Streetlights can serve not only as wireless network signal transmitters but can also be equipped with various sensors to monitor air quality and traffic, and they can even adjust lighting brightness based on environmental conditions. This isn't a vision of the future; it's the scene from last year when intelligent streetlights were introduced in Changtan Village, Keelung. Participants included not only the Industrial Technology Research Institute but also GIT, which integrated the hardware.
This is just the beginning. In GIT's blueprint, in the future, under the 5G environment, streetlights can also become tools for technology-based law enforcement. By installing cameras on streetlights that already have power sources, when they detect violations such as reckless lane changes, crossing lane boundaries, or speeding by drivers, the information can be transmitted in real-time to the cloud via the internet. After the images are analyzed by AI, automatic penalties can be issued. In the past, this required optical fiber, but with the wide bandwidth capabilities of 5G, the same effect can be achieved without the need for physical cables.
GIT with just 20 employees, excels in integration, and even before smart streetlights, they had also developed Smart Parking Meter.
In May of this year, they secured a project in which Hong Kong Telecom and French traffic ticket company Flowbird signed a contract to deploy over 10,000 Smart Parking Meter on the streets of Hong Kong in 2020. The hardware integration solution for these Meter is provided by Green Ideas Technology(GIT).
This Smart Parking Meter eliminates the need for parking attendants to record and collect fees for roadside parking. Instead, it automatically calculates parking time and payment amounts through the radar and license plate recognition system installed in the meter. It integrates with the payment backend, allowing vehicle owners to make payments directly using credit cards or mobile payment methods.
Integration may sound abstract, but it is the key to making things happen.
In order to enable communication between hardware devices, Green Innovation has developed communication technology that ensures a large number of smart streetlights won't disconnect or experience data transmission issues when Wi-Fi networks become congested. CEO of GIT, jessitsay, gives an example: when transmitting data, if one path is interfered with, it needs to automatically find the next path. If not properly designed, it's like trying to squeeze three people through a bridge that can only accommodate one person at a time, and the system will immediately encounter problems. This is based on their past experience with implementing smart streetlights on campuses, a process that involved trial and error.
In the era of 5G, these innovations from startups are enabling everyday objects around us to communicate with each other.
In the movie 'Avengers: Endgame,' each superhero is in a different location, but their 3D avatars can appear simultaneously, holding face-to-face meetings. To achieve this scene, one of the key technologies is real-time 3D video recording.
In the past, creating just 1 minute of 3D video required at least 24 to 60 cameras, a significant amount of post-production work, and took one month to produce, with costs reaching nearly one million New Taiwan Dollars. This limited its application in various fields. However, now, for 1 minute of 3D video, they only need one sensor, it takes only 1 to 2 weeks, and the cost can be kept under 30,000 New Taiwan Dollars, which is approximately only about 3% of the previous cost.
We hope that everyone can easily create 3D images," says Wu Zhe-an, the 30-year-old co-founder of Aemass, a startup specializing in 3D video systems, who studied computer graphics in the United States. He believes that 3D imaging is undeniably the most important media format of the next generation. Technologies like VR (Virtual Reality), AR (Augmented Reality), and 360-degree videos all require 3D imaging. However, the key is to lower the barriers to entry for this technology.
With the advent of the 5G era, whether through smartphones or AR glasses, the 3D images we see can be more refined and realistic. People will become more accustomed to watching videos from a 3D perspective.
Aemass's technology does not rely on traditional motion sensing but rather uses AI algorithms to replace expensive cameras and post-production processes.
Take, for example, the creation of a 3D video of Jolin Tsai dancing. They start by scanning Jolin Tsai's 3D model using a camera in a 360-degree scan. Then, using the same machine, they record her dance, even if it's only recorded from the front. AI algorithms can automatically fill in the missing details. In the future, the audience can still enjoy a 360-degree view of Jolin Tsai dancing, significantly reducing the cost and time required to produce 3D animation.
In this way, in the future, with just a smartphone equipped with a depth-sensing camera and Aemass's software, anyone can easily create 3D images.
At this stage, a U.S. bathroom fixtures manufacturer has already tested their technology to model products, allowing consumers to use their smartphones to simulate how a product would look in their space. Users can point their phones at a location, select a product, and experience how it would appear in their home. They can even freely adjust the product's size and choose different viewing angles. Moreover, in the future, this technology could be used to create models of product endorsers, enabling 3D celebrity avatars to stand next to products and demonstrate how to use them.
This is the kind of futuristic sci-fi world we used to imagine as kids," says Wu Zhe-an.