Few Shot Image Segmentation

Semantic Segmentation is one of the fundamental tasks in Computer Vision. Given an input image, the algorithm will assign a class label to each pixel. In recent years, many researchers have proposed semantic segmentation schemes that produce great performance but such schemes rely on the availability of large annotated datasets. Since data labeling is costly, researchers are now exploring few shot image segmentation schemes that allow the model to be trained using fewer images. We have started to explore this type of research. Our WACV 2021 paper provides a solution that achieves better performance than state of the arts schemes and runs fast.

Robust Autonomous Driving

Autonomous vehicles (AVs), one of the most impressive technology developed in recent years, have the potential to impact the daily lives of many people. Level 4 vehicles will include automated system which handles all driving tasks under limited driving conditionsl even if the human driver does not respond to requests to intervene while the goal of Level 5 autonomous driving is to make a vehicle perceive its environment and safely navigate any traffic situation efficiently without human intervention. However, significant technical challenges still need to be overcome before Level 4 & 5 AVs can be deployed. In this project, we intend to investigate if we can improve the accuracy of pedestrian detection under all weather conditions. We also are interested in exploring the issue of pedestrian tracking and the prediction of their future intentions. See a related NSF funded project here.  One of our research tasks is designing a fast trajectory prediction scheme. Our designed scheme was ranked first in Baidu Apollo Trajectory Prediction task for several months between Dec 2018 and June 2019.


Perceptual Understanding for Images

Encounters with manipulated and fake images are very common these days. Most of these images change the perception of reality for the audience. In this project, we take a deeper dive as to what is that change and answer questions such as what does a viewer see when they see fake images, can we identify cues that help us improve the perception of real vs. fake?



Social Psychology Inspired Facial Analytics

Predicting facial attributes and personality traits using human perception as yardstick is an important research direction in the Human-AI interaction space. Understanding what AI-based models can reliably predict, and where they have error trends, requires us to backtrack a few steps and reconnect with the principles and results of perceptual studies in the Psychological Science literature.



AI for Healthcare

AI is showcasing the transformative potential to streamline and improve therapeutic and healthcare practices. From predictive analytics to computer vision, AI technologies are breaking through traditional medical techniques to accelerate and scale multiple treatments. Our research in this area centers around “a dose of AI for healthcare data analytics”, exploring how machine learning and computer vision can advance precision diagnosis and prognosis of various diseases over different kinds of data. Some of our research interests in this area include: cognitive/computational neuroscience, machine learning/deep learning, tensor analysis, multi-modal data integration (e.g., brain imaging, imaging + genomics, mobile health data), brain network analysis, image segmentation, etc.