RESEARCH
INTERESTS

Computer vision and deep learning: video classification, activity recognition/detection;
optical flow and depth estimation; semantic segmentation; anomaly detection in
surveillance; zero-shot learning; crowd sourcing; geographic knowledge discovery.

TECHNOLOGY
SKILLS

Programming Languages: Python, MatLab, C++, Lua, Java.
Deep Learning Frameworks: PyTorch, TensorFlow, Caffe, Torch.
Others: OpenCV, AWS, Docker

EDUCATION University of California, Merced August 2014 - Expected May 2019
Ph.D. in Computer and Information Sciences. GPA: 3.9/4.0 Merced, CA
University of Kansas August 2012 - August 2014
Master of Science in Electrical Engineering. GPA: 4.0/4.0 Lawrence, KS

INDUSTRY
EXPERIENCE

Research Intern May 2018 - December 2018
Nvidia Research Santa Clara, CA
- Developed algorithms for semantic segmentation in driving scenes. Achieved
state-of-the-art performance on Cityscapes, Camvid and KITTI.
- Contributed to PyTorch implementation of FlowNet2 open sourced by Nvidia.
Research Intern Jan 2018 - May 2018
Hikvision Research Santa Clara, CA
- Developed algorithms for joint learning of optical flow, depth, camera pose etc.
from monocular videos.
Research Intern May 2017 - July 2017
TuringVideo San Mateo, CA
- Led the engineering team. Successfully delivered three products in three months.
- Developed algorithms for anomaly detection in surveillance videos including
scenarios: detecting human motion, grouping, fighting and armed.

RESEARCH
EXPERIENCE

Research Assistant August 2014 - Present
University of California, Merced Merced, CA

- Proposed a CNN architecture for real-time human action recognition and de-
tection. Obtained 10x efficiency improvements with no accuracy drop.

- Developed semi-/un- supervised approach for optical/scene flow estimation.
- Analyzed geo-referenced social multimedia including texts, images and videos
to do geographic knowledge discovery.