We have a number of funded and non-funded projects which are currently on-going.

Generation of Polygonal Geometric Art

Automatic geometric abstraction which is a process of transforming an image into geometric art has widespread use in image editing and artistic synthesis. While existing methods yield unwanted distortions, are less subject-focused and even computationally expensive, we seek to design straightforward, non-learning algorithms which can support both triangle- and polygon-based abstraction without sacrificing on the semantics of subjects in the picture.

Abby Low, Ng Ruisheng, Wong Lai Kuan, John See

Large-scale Aesthetic Evaluation of Photographs (LAttE)

Image aesthetic evaluation is a research field which aims to design computationally-driven methods which can automatically rate or predict the perceived aesthetic quality of an image or photograph by learning from image content, photographic rules and other semantic information. We investigate how features can be learned in an unsupervised manner as opposed to traditional hand-crafted rules, and design new deep learning architectures to evaluate the aesthetic beauty of photographs.

Magzhan Kairanbay, Hii Yong Lian, John See, Wong Lai Kuan

Facial Micro-Expression Analysis

A micro-expression is a brief and involuntary facial movement which reveals a genuine emotion that a person tries to hide. Psychologists have been studying facial micro-expressions since the 1960’s, computer scientists are now beginning to explore the possibility of spotting and identifying these micro-expressions using machine vision and learning algorithms; we aim to discover novel methods for doing so. This contemporary field of research has potential applications for clinical diagnosis of psychological conditions (autism and depression), criminal interrogation and lie detection. This project is in collaboration with institutions in UK and China.

Huai-Qian Khor, John See

Computational 3D Model of Forearm Rotation

The mechanism that allows for human forearm rotation is still poorly understood. As a result, sub-optimal surgical treatment of fractured radius or ulna can lead to impaired forearm motion. We aim to develop a physically accurate 3D dynamic model of human forearm rotation from CT scan, with visualization of the model, the dynamic rotation and tensing of interosseous membrane. This is a project in collaboration with National University of Singapore (NUS) and Singapore General Hospital (SGH).

Muhammad Faiz, Wong Lai Kuan, John See, Loh Yuen Peng

Past Projects

Here’s some of our previous completed projects, which are still very much relevant today.

Action Recognition in Adverse Quality Surveillance (AQuaS)

Most state-of-the-art techniques for HAR have been designed to perform well under constrained and highly controlled conditions. However, these capabilities may not be easily replicable in real-world surveillance conditions (via devices such as CCTV or web cameras) where video quality may be naturally poor. We investigate new representations for recognizing human activities in adverse quality surveillance videos.

Aesthetics-driven Stereo Retargeting and Recomposition

With the recent availability of stereoscopic displays such as 3D monitor, 3D television and stereo camera phone, there is an increasing need for stereo image retargeting and recomposition techniques. Image retargeting aims to resize an image to fit different aspect ratios and sizes while image recomposition attempts to computationally modify the composition of an image to mimic a professional photo. We investigate new aesthetic-driven methods for retargeting and recomposition for stereo image pairs.

Long-term Video Surveillance (LoViS)

In a long-term period, video surveillance takes on a different perspective. Habitual behaviors of people or permanent changes to objects can be observed while anomalous “out-of-norm” variations can also be traced. We investigate how these variational patterns can be extracted over a long period of time to gain a high-level understanding of various factors at play.


External (Industry-funded)

External (Government/Institutional-funded)