LSE - Small Logo
LSE - Small Logo

Takahiro Yamamoto

April 11th, 2025

Facial Recognition and the Two Expos in Osaka

0 comments | 1 shares

Estimated reading time: 10 minutes

Takahiro Yamamoto

April 11th, 2025

Facial Recognition and the Two Expos in Osaka

0 comments | 1 shares

Estimated reading time: 10 minutes

A visitor to Osaka in western Japan this month may feel that the city is making strides toward the future through the use of biometrics. The Osaka Expo, due to open on 13 April 2025 under the theme of “Designing Future Society for Our Lives,” will use facial recognition to manage entries and exits of multi-entry pass holders and staff. This follows the Tokyo Olympics—held without admitting spectators amid the Covid-19 pandemic in 2021—that managed the movements of athletes and staff with the same technology. Now the Osaka Expo will broaden the scope of facial recognition to include visitors, which are expected to be in the millions over the course of the event’s six-month duration. Outside the event venue, since late March, 130 out of 134 stations of the Osaka City Metro have introduced facial-recognition-based ticket gates to passengers who choose to pre-register their personal information and payment methods (shown below). There is nothing particularly new about the technology itself, as it has already been introduced by other railways in Japan. It is already in wide use on high-speed railways in China, metro stations in Moscow, and at boarding gates and immigration checkpoints in airports around the world. But these two usages of facial recognition indicate that the whole city of Osaka is pushing toward the integration of facial-recognition technology into people’s lives.

A facial-recognition-based ticket gate at a station in Osaka City Metro in March 2024, then at the pilot phase. Photograph by the author.

It will be Osaka’s second time hosting the World Expo. The 1970 Osaka Expo was the first one to happen outside of Europe and North America. With its theme of “Harmony and Progress for Mankind,” the 1970 Osaka Expo provided a stage for postwar Japan to impress the world with its economic growth and societal development. One of the 116 pavilions there, built by a Japanese conglomerate Sumitomo, included an exhibit entitled “The Computer Physiognomy,” an early effort for pattern recognition of human faces by machines. A participant to this popular exhibit would sit in front of a camera and get their facial portrait taken. The computer then analysed its features and classified it into one of the 21 pre-registered types, each represented by a Japanese or international celebrity, including Marylin Monroe, John F. Kennedy, Mahatma Gandhi, and Saigō Takamori (the model for the film The Last Samurai). That verdict of likeness would come with the computer’s judgement about the participant’s personality and occupational aptitude. The exhibit thus presented a vision for the future, where machines will become so intelligent as to recognise people’s faces and predict what will happen to them.

The Computer Physiognomy ran on a mainframe computer called NEAC2200, the most advanced from Nippon Electric Company (NEC) at the time. NEC has been a pioneer in image pattern recognition research, constantly achieving a top rank in standardised tests for accuracy of facial recognition since the late 2000s. Its products have been exported to dozens of countries, including India, where its national ID system Aadhaar that manages over one billion people runs on NEC’s system.

This same company, NEC, provides the facial recognition system utilized in the 2025 Expo. Takahashi Atsushi, NEC’s engineer responsible for the 2025 Osaka Expo, referring to the company’s history of providing facial recognition for Expos, noted that  “it gave me a renewed sense that NEC is where it is today because it values technologies and relentlessly pursues their refinement.”

There is a degree of irony in the use of Expo as an opportunity to develop facial recognition. Since its beginning in 1851 in London, international expositions have been a venue where empires set up booths featuring their colonies to flaunt their political, economic, and cultural power. Japan was no exception. For instance, in 1904, at the Louisiana Purchase Exposition in St. Louis, Missouri, it placed nine indigenous Ainu people on live display.

Throughout the 20th century, colonial officers and scientists traveling across Asia, Africa, and the Pacific photographed, analysed, defined, classified, and persecuted the bodies of others, people of “suspicious traits,” using biometric data such as faces, body measurements, and fingerprints. In this context, the act of creating people’s standardised portraits has been inextricably linked to the violence of twentieth-century empires, as it provided the ideological and material basis for race science, criminal investigations, and ethno-racial politics. As Sano Mayuko has pointed out, it was not until 1962, in Seattle, that colonial exhibits disappeared from the Expo.

The technique behind the Computer Physiognomy was derived from the field of quantitative anthropology, which was founded on principles of prewar eugenics. Yamazaki Kiyoshi, a dentist and anthropologist who participated in The Computer Physiognomy project, had studied craniometry in France in the 1920s and had devoted his research to the use of quantitative methods to analyse the human face. Even after the Second World War, when eugenics was rejected and discredited, Yamazaki pursued what he called “dental approaches” to human faces, which avoided inter-racial comparisons but still aimed to study the Japanese race through the quantitative research of skulls.

Research and development of image pattern recognition, including that of human faces, had been underway elsewhere in the 1960s, including in the United States where the Central Intelligence Agency covertly funded scientists, as a recent study shows. To the best of my knowledge, however, there was no equivalent of the Computer Physiognomy that made this nascent technology publicly visible and allowed spectators to experience it, albeit in a rudimentary manner. Thus, the Computer Physiognomy at the 1970 Osaka Expo needs to be understood as one of the important origins in the history of facial recognition technology.

Today, facial recognition software can identify individuals in milliseconds and assess their age, gender, or ethnic background. Reaching this level of sophistication would take several decades of engineering research, a rise of neural network artificial intelligence, and a rogue engineer. As Kashimir Hill and others have demonstrated, after the technology became sufficiently advanced, it was not a large tech firm (which worried about privacy) but a then-little-known entrepreneur who built a “Google search for faces” and began selling it to law enforcement offices in the United States in 2019. This software, Clearview (now called Clearview AI), was matching faces with millions of facial images recklessly scraped from the Internet without consent from the individuals or institutional platforms.

When compared to the above, the Computer Physiognomy’s algorithm was incredibly simple—it was based on just around 3,000 photographs which Yamazaki manually clipped from newspapers, and its personality assessment used just three parameters taken from the participant’s facial image. Yet what Yamazaki and his colleagues accomplished in 1970, as colonialism wound down (at least at the level of national government), was to prolong the life of quantitative analysis of human faces. This also masked its power with the alleged objectivity of the machines and the allure of the ability to predict a future.

Fifty-five years after the first Osaka Expo, the public rollout of facial recognition technology is in full swing, even with the common knowledge of its racial biases and oppressive nature. The 2025 Osaka Expo has no qualms about subjecting the public to the machine’s statistical gaze that had once required extensive rhetorical cover to dissociate itself from colonial governance. The mega event that parted with colonialism in the late 20th century now manages attendees with the exact method that sprang out of imperial biopolitics. In 2025, amid broader economic and societal struggles, Osaka is hosting the Expo in an attempt to repeat the “celebration capitalism” to boost its economy and pride (a failing effort, as Saitō Kohei notes). The rise of facial recognition shows, however, that long after Expo stopped being the showcase of empires, that the city is championing the comeback of the hierarchical gaze to the public, oblivious to the technique’s—and the country’s—colonial baggage.

About the author

Takahiro Yamamoto

Takahiro Yamamoto, a historian focused on modern East Asia and the Pacific, is a lecturer in the Humanities, Arts and Social Sciences (HASS) cluster at Singapore University of Technology and Design. His publications include a monograph Demarcating Japan: Islanders, Imperialism, and Mobility, 1855-1884 (Harvard University Asia Center, 2023) and an edited volume Documenting Mobility in the Japanese Empire and Beyond (Palgrave Macmillan, 2022).

Posted In: Asia | Contemporary | Global | History | History of Technology | Imperial History | International History

Leave a Reply

Your email address will not be published.