Remembering Sung Kah Kay
Updated: Oct 31, 2022
The next time you use your digital camera, please remember my friend, Dr. Sung Kah Kay.
Dr. Sung, Singaporean son, was the first person in the world to create an effective algorithm for face detection. This is the ability to analyze a digital image to determine which pixels form a human face, and is the reason faces appear nice and sharp in your photos. He invented this in 1996 for his PhD thesis at MIT, but it was only about nine years later that digital cameras started to incorporate face detection. Today, it is a standard feature that is taken for granted. Some cameras draw a box around every face that is detected, showing this feature in action.
By demonstrating that a machine learning algorithm could find faces of different sizes at any position in an image, a feat far from obvious back then, Dr. Sung’s invention started a gold rush into other kinds of facial image analysis. Many apps today, such as Apple’s FaceID (facial recognition), CheckMyAge (age estimation), SnapChat filters (face masks), and ModiFace (virtual makeup), are inspired by face detection, and require face detection as an important first step.
It would not be hyperbole to say that Dr. Sung’s work has made a big contribution to automatic image analysis.
Seeing the far-reaching impact of one’s creation, many inventors would have bragged about it. But if I know him well enough, Dr. Sung would prefer to remain unsung. I first met him in 1989 when he returned to MIT to start his post-graduate studies after completing National Service. I was then in my final year as an undergrad. Together with other students in the Singapore Student Society (SSS), we often met for meals, games, and outings. And although he came from a well-to-do family, Kah Kay never had any airs, nor was he aloof. On the contrary, he was sociable, well-liked, and smart.
Very smart. He would eventually earn all four levels of degrees that MIT awarded: Bachelor, Master, Engineer, and PhD.
I recall going head-to-head with him for the inaugural SSS Nerd Cup. This was a fun competition to see who was “more nerdy”. (Let the reader understand that MIT students take pride in being nerds. We routinely solve calculus problems in our sleep, and derive satisfaction from doing so.) I also recall threatening to spam my friends’ mailboxes with constant reminders to vote for me to win, while Kah Kay solicited votes using his charm and wit. He won. As nerds go, he clearly had better EQ than I did.
Years later, our paths would cross again when we both joined NUS in 1996. I was Senior Tutor while Dr. Sung was Assistant Professor. He was still the same self-effacing and fun-loving guy that I knew. And he still drove the same car: Acura Integra in the US, Honda Integra in Singapore. Black. We would often talk about machine learning and about pedagogy. He was very prescient when he remarked that the big challenge in machine learning was to incorporate “expectation” into an AI model. After all, humans learn very well because we use expectation to guide our interpretation of sensory signals; that is, we literally see what we expect to see. This is a fact well established by neuro-psychologists. Likewise, machine learners should greatly benefit from using expectation too. Alas, this challenge still remains unsolved today.
Dr. Sung was also an excellent teacher, winning a teaching award in just his first year of lecturing. When I asked for his secret, he told me to always put myself in the students’ shoes, and to understand how my lesson is being received. He would go on to win the teaching award again in subsequent years. And so did I, when I heeded his advice.
Tragically, in a perverse twist of fate, Dr. Sung was killed when his aircraft, flight SQ 006, crashed at Taipei’s airport on the typhoon-battered night of 31st October 2000. Exactly twenty-two years ago today. He was only 35 years old. His newly-wedded wife, Jennifer Loo, also perished with him on that same flight. I reacted to the news first in disbelief, then in shock, and then with great sadness. Many in NUS, students and staff alike, expressed with tears what words could not. We had lost a collegial colleague, a role-model researcher, and an inspirational instructor.
So, the next time you use your digital camera, please remember my friend, Dr. Sung Kah Kay.