Embrace the upgrading of the manufacturing industry and focus on machine vision – An interview with Gpixel

The CMOS image sensor in machine vision is expected to experience double-digital growth over the next five years as factory and logistic automation grows, according to Yole Développement’s latest report, Machine Vision for Industrial and Automation 2021. Sales of CMOS image sensors for machine vision could grow from $266 million in 2020 to $527 million in 2026, representing a 12.1% compound annual growth rate. Industrial cameras could also see the emergence of a 3D ToF sensor or the integration of ever more high-resolution, high-frame-rate CMOS image sensors.

In this context, Richard Liu and Florian Domengie, both Analysts in the Imaging team within the Photonic and Sensing Division at Yole Développement, had the opportunity to connect with Wim Wuyts, Chief Commercial Officer at Gpixel. He told us about the company’s products, strategy, and prospects for industrial cameras. Read their discussion below.

Richard Liu (RL): First, could you introduce yourself and Gpixel to our readers?

Wim Wuyt (WW): I’m Wim Wuyts, the Chief Commercial Officer of Gpixel. Gpixel is a fabless company providing image sensors to the specialty markets, including scientific and medical imaging, industrial inspection and automation, and professional imaging, including photography, cinematography, and high-speed video.

RL: Please briefly introduce the core team of Gpixel and the division between Europe and China?

WW: Gpixel was founded in 2012 by experienced CMOS image sensor designers and semiconductor physicists, including our CEO, Xinyang Wang, who worked as an image sensor designer in Belgium before returning to China to establish the company. We are headquartered in Changchun, China but also have sites in Hangzhou, Antwerp, and Tokyo. All sites include design and marketing staff. The company’s testing cleanrooms, supply chain staff, customer service staff, and shipping facilities are in Changchun. A lead design center is selected for each new product concept, and there is collaboration between sites for product management, test development, and applications support.

RL: What new products have Gpixel developed recently? Could you please expand on the applications you are targeting for these new products?

WW: Today Gpixel’s portfolio includes 6 main product families. The GSENSE line of scientific CMOS imagers includes rolling-shutter backside-illuminated sensors with very low noise and are popular in microscopy and astronomy. The most recent addition to this family is GSENSE4040BSI, a 16 MP backside-illuminated sensor, providing an upgrade path from the discontinued CCDs used in astronomy and similar applications. The new GLUX sensor family is based on the same scientific imaging technologies but offers very large pixels size and MIPI interface for easy integration, targeting the lowest light applications, like surveillance under starlight conditions. The GMAX sensors offer global shutter, high frame rates, and resolutions starting from 5MP up to 100+ MP and are used in applications ranging from 3D laser scanning to aerial imaging. Our new GSPRINT family also offers global shutter pixels, but with larger pixel sizes and higher speeds, for example, the GSPRINT4521 provides 21MP resolution at 1000 fps. Gpixel also provides high-speed linear sensors in a variety of resolutions and recently launched the GL3504, which provides both 2k and 4k resolutions in a C-mount compatible optical format as well as the GLT5009BSI, the first in a family of backside-illuminated TDI sensors. Gpixel also recently introduced a new product family with the GTOF0503 indirect time of flight sensor for depth sensing applications.


RL: Regarding the iToF sensor for 3D imaging applications, could you please tell us more details about the specification of this sensor?

WW: The GTOF0503 sensor is manufactured using the pixel-level stacking and BSI technologies offered by Tower Semiconductor’s 65 nm process and features a VGA array of 5? m 3-tap state-of-the-art iToF pixels. The GTOF0503 sensor is attractive for industrial applications due to the combination of a small, cost-effective footprint with industry-leading depth accuracy and precision at short-, mid-, and long-range. This allows for exceptional depth sensing even in challenging ambient light conditions using the pulse modulation iToF technique. A demodulation contrast of > 80% is achieved with modulation frequencies of up to 165 MHz at either 60 depth frames per second in Single Modulation Frequency mode or 30 fps in Dual Modulation Frequency mode.


Florian Domengie (FD): For 3D imaging and sensing in machine vision, there are several types of approaches, such as structured light, laser triangulation, stereo, and ToF. How do you see the 3D machine vision market evolve? Do you think ToF will become one of dominant technology in the industrial market.

WW: There are a variety of requirements for 3D industrial imaging, and each technique has advantages and disadvantages for specific applications. Our aerial imaging customers are unlikely to adopt iToF as a replacement for photogrammetry in generating 3D models of buildings from airplanes, for example, because of the large imaging distance. Our high-resolution GMAX sensors are often used in systems that employ structured light, and we don’t expect iToF systems to replace this technique for applications where high resolution is the main requirement and the ambient light is controlled. Our lower resolution GMAX sensors are popular in laser scanning systems because of their high speed. Laser scanning is very adaptable to specific tasks and has the advantage of also providing distance measurement but is primarily useful where low-resolution scans at high speed match the task. As the resolution available in industrial time-of-flight systems increases, it will become an even more compelling alternative to laser scanning, especially in vision-guided robotics, bin picking, automated guided vehicles, and factory automation.

FD: The mobile camera market is quite sensitive to the cost of the CIS sensor. How sensitive is the machine vision market to the sensor price?

WW: All image sensor applications have cost constraints, but certainly, in the industrial space the pressures are different than in a mobile phone. The purpose of installing an industrial camera is generally to save money by throughput increases, yield improvements, and quality enhancements. The first job of the image sensor supplier is to provide performance that will enable these cost savings by the end-user. Within this context, there is an inherent value to the system, and in the end, the price must match that value to deliver an attractive return on investment. How all these factors play out are very application-specific.

RL: In mobile CIS CMOS Image Sensor, the race for pixel resolution is ongoing, but we see that the high-end mobile wants a bigger format CIS for better photography. What is the trend for CIS in machine vision?

WW: Reducing pixel size has been a general trend in machine vision for decades. Thanks to improvements in noise, pixel charge capacity, and electronic shutter efficiency, tasks that used to be done with 7.4 um CCD pixels are handled easily today by the 2.5 um global shutter pixels in Gpixel’s GMAX product family. But pressures are slowing this trend, especially in very high-resolution applications where it’s expensive to develop optics compatible with the smallest pixels, and either higher dynamic range or signal-to-noise is required. For this reason, many of Gpixel’s highest resolution GMAX products incorporate a larger 3.2 um pixel.


RL: In the field of high-end machine vision, CIS has high-level specifications but with high value, which is also the field of some traditional manufacturers. Does Gpixel have any plans to enter these fields, such as Aerospace?

WW: Since the company’s founding, Gpixel has focused on some of the most demanding imaging tasks, initially in scientific imaging. Gpixel was the first company to offer BSI technology on CMOS sensors for scientific imaging, and today you can purchase scientific sensors from Gpixel with quantum efficiencies over 95%, noise below 1 e-, and size up to 90 mm square. In the last several years, we have successfully leveraged the collective experience of our growing team to expand into some of the most high-performance applications in industrial imaging. Gpixel doesn’t specifically target space applications, but we will continue to look for opportunities to put our high-performance technologies to work, both in our standard product portfolio and in our custom developments. Being active in scientific and specialty markets doesn’t mean we are active in defense applications though. As a strategic company guideline, we avoid developing and/or offering products for military end-use. Being an international company with several office locations across the world and a global supply and end-customer base, we need to steer away from any restrictions that are connected to the military business.

RL: In recent years, Gpixel quickly capturing business opportunities in machine vision even though the market has strong and established competitors. Could you please share the “secret to success” behind this? And what do you think is the position of Gpixel within the industry imaging ecosystem?

WW: One of Gpixel’s key advantages is the ability to incorporate new technologies very quickly, so our sensor portfolio offers customers early access to new performance improvements in image sensors. For example, Gpixel was the first to offer backside illumination in CMOS sensors for scientific use, the first to offer industrial sensors utilizing a 2.5 um global shutter pixel, and today, GSPRINT4521 is the fastest CMOS sensor available as a standard product, offering 21 MP at 1000 fps. The new time-of-flight sensor is based on both backside illumination and pixel-level wafer stacking technology. This technology leadership position, combined with Gpixel’s competitive pricing, makes the portfolio especially compelling. In addition, Gpixel has a high bandwidth for product development and an advantage in time-to-market over the traditional players. Especially in the last few years, some industrial sensor suppliers have been slow to introduce new products, providing a window of opportunity for Gpixel. At the same time, the Gpixel team has been growing to include many veterans of industrial imaging in both design and marketing, making the path to seize the opportunity all the more clear. Today, Gpixel is the preferred supplier for specialized, higher-end industrial applications and, at the same time, a growing participant in the mainstream of the market.


FD: Gpixel is now serving several main markets, such as Europe and China. One is a developed market, and the other is a developing market; both are in great demand. Could you please provide your insight into these two markets?

WW: Today just over half of Gpixel’s revenue comes from sales in China, with the remainder split between Europe, North America, and the rest of Asia. Although China may be a developing market for automation systems, it is a mature market for industrial image sensors. The industrial image sensor and camera markets have been globalized for many years, with camera makers from all continents sharing similar requirements and focusing on sales across the globe. As a result, the camera makers who use our sensors tend to focus on specific subsegments of the market more than on specific geographies. For this reason, the location of our camera partners (and therefore Gpixel sales revenue) doesn’t reflect very accurately where the end-user might be.

RL: Any other message you would like to share with our readers and the industry?

WW: Only to thank Yole for giving us the opportunity to share information about our products and our view of the market with your readers. We appreciate the work done by Yole to facilitate information sharing that benefits all market participants.


Wim Wuyts received the Master degree in Electronical Engineering from the University of Antwerp, Belgium in 2002 and is a Senior Marketing, Sales & Business Development Executive with 19 years working experience in sales, marketing, business development and general management functions in CMOS Image Sensor companies.

Starting as Product Manager with FillFactory in 2002 where he was responsible for the Application Support of the standard product portfolio. After the acquisition by Cypress Semiconductor in 2004 Wim started from 2006 in the Business Development group dealing with custom NRE project designs of CMOS image sensor products. With the acquisition of the CIS activities of Cypress Semiconductor by On Semiconductor in 2011 Wim also managed the CIS application support team. Early 2012 Wim was hired by CMOSIS to become Director Sales and Marketing to help growing the company to a world leader in specialty CMOS image sensor products where he was deeply involved in growing start up business from €10 million to €90+ million annual revenues. After the acquisition of CMOSIS by ams in 2015 Wim left ams in August 2018 to co-found Gpixel NV. Today Wim is in charge of the global marketing and sales activities of Gpixel as Chief Commercial Officer.


Richard Liu is a Technology and Market Analyst in the Photonics, Sensing & Display division at Yole Développement, part of Yole Group of Companies. Based in Shenzhen (China), Richard is dedicated on imaging activity (Monitors) as well as the development of technology & market reports
Prior to Yole, Richard was engaged in camera module design on image sensor, AF/OIS at Onsemi, before this, he worked as a customer-application-technologist in Micron/Aptina Imaging. Richard has over 12 years post graduate experience in both of imaging semiconductor and camera module industry, he has the successful track record in developing projects for the tier one smart phone and module makers, which brought him wide industry connection in the CMOS image sensor supply chain and ecosystem
Richard graduated from Wuhan University (China) and holds an Electronics Engineering Degree.

Florian Domengie, Ph.D., is a Senior Technology and Market Analyst, Imaging and Display within the Photonics & Sensing division at Yole Développement (Yole). Florian is engaged in technology and market analyses for various imaging technologies and contributes to the production of relevant reports and projects. Prior to Yole, Florian worked in STMicroelectronics in process and technology development and R&D project management.
Florian has authored and co-authored numerous papers and five patents in semiconductor R&D and manufacturing.
Florian holds an M.Sc. in Engineering Physics, Materials and Microelectronics from INSA Toulouse (France), and a Ph.D. In Microelectronics and Nanoelectronics from the University of Grenoble-Alpes (France).

Related Report

YINTR21135-Machine Vision for Industry and Automation 2021_bd

Machine Vision for Industry and Automation 2021
The industrial vision market matters, and the ecosystem is reconfiguring.