Taiwan may already be a key part of the electronics manufacturing supply chain, but it is now trying to enhance its image beyond manufacturing, as an enabler of hardware-based artificial intelligence (AI) services serving an increasingly data-driven society.

This was evident when we recently spoke to the country’s science and technology minister Liang-Gee Chen during Innovex, a startup exhibition that ran alongside Computex. In addition to the AI services focus, he also emphasized a desire to connect its startups with global innovation ecosystems, provide them with a platform for growth, and encourage its academics to spinout out technologies that enable an AI enabled world.

RaviBelani

Ravi Belani (source: Ravi Belani)

One of the pillars of the government’s master plan is to encourage key startup accelerators to establish in Taiwan or recruit for its cohorts from the country – an important part of the global ecosystem integration plan. A number of accelerators and investors from the U.S. were invited to meet local startups during Innovex. We had the opportunity to meet some of those investors while in Taipei.

One of them was Ravi Belani, managing director of the Alchemist Accelerator, a venture-backed initiative based in Menlo Park, Calif., dedicated to accelerating business-to-business focused startups. He told us he first received an invitation to visit Taiwan two years ago. “After that trip, I loved it. It has the best of Japan (in terms of efficiency and work ethic) and China (in terms of entrepreneurial characteristics).” He added the Taiwanese are incredibly smart but not arrogant. “They are not letting their intelligence get to their head.”

He makes strong arguments for his new interest in Taiwanese entrepreneurs and why he’ll be looking more closely at the country. “Japanese founders don’t think necessarily outside Japan. And in South Korea, there is entrepreneurial energy but not great thinking globally. Also, Taiwan is a really good gateway for China and the U.S.”

He added, “Taiwan is really ahead of Silicon Valley in terms of hardware — and if you own the hardware, you own the platform.” Belani said companies are now increasingly becoming hardware companies, and Silicon Valley hasn’t continued to build the expertise in this area the way Taiwan has. “That’s why I’m very bullish about Taiwan.” In fact, he has two Taiwanese startups currently in the its accelerator program, CoolSo and PenguinSmart.

New HMI enables gesture control
One of the co-founders of CoolSo is Jack Wu, previously involved in medical device development for eight years at Taiwan’s research institute ITRI. Other founders have backgrounds in research and engineering at National Taiwan University and Mediatek. The startup, founded in 2017, claims to provide an innovative new approach to gesture control. Based on an Arm Cortex M processor and high sensitivity motion sensor, it has patented a technology that measures individual muscle activity to recognize both gestures and strength levels.

CoolSo

CoolSo's gesture recognition and control technology measures individual muscle activity
to recognize both gestures and strength levels. (Image: CoolSo)

The startup claims that unlike other human machine interface (HMI) solutions that respond to gestures, its solution is immune to environmental interference, can be used indoors and outdoors, and is compact enough to be integrated into a wearable or other device for human body movement measurement and analysis. This contrasts with image processing-based solutions that can be affected by outdoor conditions and sunlight, or others that rely on electrical signals where the sensors and modules are still not compact enough for integration into any electronics.

Blood pressure measurement using facial recognition

Continuing on the biomedical theme, another of Taiwan’s startups that emerged from a research project and with initial seed funding from the government’s program to help professors commercialize their research is FaceHeart. We met up with the its chief strategy officer Jerry Chang, who recently returned to Taiwan from Silicon Valley, at the company’s office in Hsinchu City.

FaceHeart is targeting mainly medical as well as security applications — it can read heart rate variability (HRV) and blood pressure using facial images, with almost 98% accuracy, according to Chang. The company has just received series A investment from Mediatek. Although the figure was not disclosed, Chang hinted to us that it was double-digital millions in US dollars, and the official announcement values FaceHeart post investment at NT$500 million (around US$16 million).

Developed by a team led by Prof. Bing-Fei Wu from the National Chiao Tung University (NCTU), it uses commercially available cameras to measure users' heart related parameters just by ‘looking at’ users' faces and without them having to be interrupted. The readings are processed on Nvidia-based systems which first carry out face detection and then apply digital signal processing to determine the vital measurements such as heartbeat rates and blood pressure.

FaceHeart

FaceHeart demos its boold pressure measurement using only facial recognition
(Image: FaceHeart)

The reading accuracy can be high as 2 bpm (beats per minute) to 3 bpm when the patient is in still mode, which FaceHeart says fulfills medical requirements. Even when the patient is running on a treadmill, the reading accuracy remains within the range of application. The company adds that no high-resolution camera is required — a 30 fps camera is good enough for measuring heartbeat rates.

The system, including hardware and software, was developed by the team from the NCTU while the particular algorithm chip was developed through a partnership with MediaTek. The challenges in developing the system included figuring out how to precisely read signals in dim light at night or in a dark room.  Under such circumstances, it’s usually necessary to adjust the aperture or set the shutter speed of camera. But, the NCTU team eliminated this factor by putting this task on the AI, which took five years to overcome before coming to the final product.

Chang said the idea is to enable contactless heart measurements, which is a big opportunity in remote telecare, especially in countries like China. “China is facing an elderly people crisis more than ever. The majority of major diseases don’t need a hospital visit. They want our camera system for this, the televideo.” The system is targeting the capability to read heart rate, heart rate variability, blood pressure and blood oxygen using facial images, which can also be interpreted over remote video as is needed for remote telecare.

Chang added that the company is also targeting the U.S., probably by the end of this year, for high-end nursing homes and other medical applications. FaceHeart already has revenue coming in from a security application at Shanghai Commercial and Savings Bank in Taiwan. In the latter, its system can use facial recognition to monitor customers in high-value or high-risk transactions to identify potential fraud based on the customer’s vital measurements. Chang said another use is in courthouses, where lie detection could be measured based on contactless heart rate measurements obtained through facial recognition.

Shrinking AI into the edge to make video more useful
One of the other high-profile AI startups we met is OmniEyes. We talked to one of the three professor co-founders, Chun-Ting Chou, about what they are doing that’s different, success to date, target markets and future plans.

The founders have a background in wireless networking, multimedia and machine learning. They brought this together under the government’s program that funds professors to commercialize their research, and set up OmniEyes, which now has nine people working for it. The firm collects video data from dashcams and commercial-grade live street recordings to provide useful location-based information for real-time task execution.

In Taiwan, almost every car has a dashcam fitted, mainly to provide evidence under motor insurance claims. Chou said, “These collect so much information about the city, environment, gas prices, points of interest and so on, but most of the information is buried in the video. We are using AI to automatically interpret the information.”

Their intention is then to sell this ‘actionable’ information to relevant groups of users; fleet management is an initial target sector. Chou said some of the challenges for fleet managers include the cost of vehicles idling, and the high potential for traffic violations as drivers face greater pressure to deliver within ever shorter timelines. OmniEyes delivers a turnkey package for fleet managers that can help address these issues and optimize fleet operations.

OmniEyes

OmniEyes detects objects, points of interest and other data buried in video images from dashcams (Image: OmniEyes)

We asked how what they are doing is different to the way companies like MobileEye are offering their service. Chou said that MobileEye is using data only for closed loop systems, and not collecting video data for other purposes. He added that OmniEyes’ AI engine sits in the on-board unit (OBU) in the dashcam, and fleet managers pay monthly or annual subscriptions per vehicle to get key parameters or information extracted from video images that OmniEyes collects on its platform (which collects video not just from that fleet but multiple fleets and other live street video sources).

He explained the challenge was in analyzing the videos in the OBU itself, shrinking large convolutional neural networks (CNNs) onto a small lightweight device without losing performance. “That’s what we do — we shrink the AI into the end device, and then move it to the cloud. Shrinking is the OmniEyes secret sauce.” The closed-loop system sends pseudo-labelled data from the end device to the cloud, where the data is used for training, and then shrunk, and sent back to the device OTA (over-the-air) interface.

On the issue of data privacy, Chou stressed that OmniEyes is not carrying out facial or people recognition, nor does it release road pictures. The closest they get to user-specific data is identifying license plates, but they don’t connect this back to driver’s identity. On the latter, he said they are already talking to potential customers in the U.S. interested in using the license plate detection capability.

Plugging into global networks for better chance of success
The above are a tiny selection of the startups we were introduced to in Taipei, and it certainly feels like a vibrant ecosystem that is keen to grow. The highly interconnected networks of international and local incubators and accelerators ignited by ambitious government programs to commercialize research is certainly creating a strong pipeline of hardware-based AI services companies in everything from health tech and industrial tech to agri-tech. Together with a growing army of local and foreign venture investors keen to bring some of these startups into their portfolios, this will definitely give Minister Liang-Gee Chen some of the outcomes that he’d envisioned, over the medium to long term.