The “Hong Kong Early Tabloid Newspapers”《香港早期小報》, launched in 2022, has collected tabloid newspapers published in Hong Kong during the 20th century. Contrary to the serious broadsheet newspapers, the tabloid newspapers represents the leisure and entertainment of the masses, covering topics like politics, operas, dramas, comics, and pornographies etc.

Previously, we have explored in text analysis and visualization on The Observatory Review, one of the politically oriented tabloid newspaper from the collection. However, the text analysis has only showed a glimpse of our tabloid newspaper collection. To further promote its multifaceted content and improve its accessibility, we want to look into another tabloid newspaper with a different theme and focus on the newspaper’s illustrations.

To do so, it would be essential to recognize and extract images from the tabloid newspaper. Fortunately, we came across a visual analysis tool built for such task at the Digital Humanities 2022 Tokyo Conference (DH 2022). At the Visual Analysis of Printed Illustrations using Computer Vision workshop, researchers from University of Oxford presented a newly developed processing pipeline that applied machine learning to detect, match, compare, and classify printed illustration. The illustration detector python package is easily accessible and allow user to retrain the existing model. 

In addition, publication of digitized Latin script newspapers have proceeded in recent years, but new publication in Asian languages is still rarely seen. We hope to contribute by applying existing optical character recognition (OCR) tool to our Traditional Chinese tabloid newspaper. 

Project goal

In this project, we aimed to develop a pipeline to automatically recognize, extract, categorize, and display illustrations from the selected tabloid newspaper. We tried to explore tools from academic Computer Vision repository, business OCR service, and opensource library and framework. Hopefully, in the future, we might be able to apply the streamlined process to another newspaper or other digitized collections. 

Why Amusement News is selected?

Amusement News《娛樂之音》is selected for its theme in entertainment, especially movie and Cantonese Opera oriented content, and its richness in printed illustrations. 

Cantonese Opera, often known as the great theatre or Daai Hei (大戲), represents one of the traditional xiqu (戲曲) genres derived from Guangdong (廣東). It is a recognized traditional culture which has been inscribed on The National List of Intangible Cultural Heritage of China and United Nations Educational, Scientific and Cultural Organization’s (UNESCO) Representative List of the Intangible Cultural Heritage of Humanity. Cantonese Opera also represents the long-standing popular entertainment enjoyed by the locals in Hong Kong. During festivals and temple fairs, like Luna New Year and Birthday of Tin Hau (天后) or Guanyin (觀音), people go and watch Cantonese Opera performed at ad-hoc Bamboo Theatre (戲棚) in “big piece of land” called Tai Tat Tei (大笪地). 

Since the late 1940s where Shanghai film companies moved to Hong Kong, movie and cinema started to thrive. With the inflow of capital and talent, Hong Kong soon turned into a movie-making hub and started gaining international recognition gradually. As more and more cinemas with affordable pricing established, movie watching became one of the most popular entertainment among the masses. 

The Cantonese Opera industry caught on to the flourishing movie-watching culture and transformed Cantonese Opera performance into films. During the 1950s, Cantonese Opera was one of the most prolific generas in cinemas where over 500 Cantonese opera movies were produced. Later, Cantonese opera also turned into videotapes, fanzines, and CDs. 

Among the many tabloid newspapers, the Amusement News reported extensively about cinemas, theatres, and Cantonese Opera, covering topics about the industry, the scripts, and the stars. On each issue, there were photos captured from live performance, movie, and behind the scenes. The Amusement News also involved in influencing the popularity of movie and Cantonese Opera. For years, they organized the “Three Champions of the Opera World” (梨園三王) contest for audience to vote for the best male lead “King of Wenwusheng” (文武生王), the best female lead “Queen of Huadan” (花旦王), and the best male clown “King of Chousheng” (丑生王) in Cantonese Opera of the year. 

Our Amusement News data set covered issues from 1952 to 1959 which captured the development of movie and Cantonese opera at the time.


Fonoroff, Paul. Silver Light : a Pictorial History of Hong Kong Cinema 1920-1970. Joint publishing H.K. Co. LTD., 1997.

岳清. 花月總留痕 : 香港粵劇回眸1930s-1970s = Remembrance of evanescent times past : a retrospective look at Hong Kong Cantonese opera. 香港第一版., 三聯書店香港有限公司, 2019.

Heritage and Integration- A Study of Hong Kong Cantonese Opera Films (