Selected Projects

• 16-824 Visual Learning and Recognition, Spring 2021, Carnegie Mellon University
• Other members: Alex Hung, Muyu Ouyang

Page Github

We explore using a U-transformer and multiple combinations of the model structures to track the inserted needle in ultrasoung images.

• 16-833 Robot Localization and Mapping, Fall 2020, Carnegie Mellon University
• Other members: Qindong Zhang, Ishraq Bhuiyan

Github

We use ICP-based SLAM, HDL and Kalman filter to localize a ship welding robot in a GPS-denied environment.

• 11-685 Introduction to Deep Learning, Fall 2020, Carnegie Mellon University
• Other members: Ashley Kim, Ran Zhang, Jianchang Huang

Github Presentation Report

We explore transfer learning and active learning methods to detect the pedestrian in the thermal images using faster R-CNN and YOLO.

• 16-725 Medical Image Analysis, Spring 2020, Carnegie Mellon University

I implemented a classical speckle reduction anisotropic diffusion (SRAD) filter in python, and improved the filter performance by adding histogram matching and edge preservation. I added histogram matching after several iterations to keep the contrast, and modified the original diffusion coefficient based on the result of Canny edge detector. I found that the new SRAD filter performs better in noise reduction and edge preservation, and have higher SSIM and Figure of Merit (FOM).

• 16-741 Mechanics of Manipulation, Fall 2019, Carnegie Mellon University

I implemented an optimization-based planning method trying to minimize a goal function defined by the torque and the length of the trajectory. The project explores how different optimization goals influence the optimal trajectory.