à£à£Ö±²¥Ðã

The 3D Perception and Scientific Imaging Laboratory advances 3D perception research across both natural images and specialized scientific imaging modalities. From everyday scenes captured by commercial cameras to complex structures revealed through advanced microscopes and telescopes, our goal is to develop efficient computer vision algorithms to unlock deep understanding at both macroscopic and microscopic scales.

Research Focus Areas

• Image-based object pose estimation with deep geometric understanding
• 3D segmentation, reconstruction and analysis from specialized scientific imaging modalities
• Interdisciplinary applications of 3D vision algorithms for assistive systems

Current or Key Projects

1. Object pose estimation from natural images: The goal of this project is to develop efficient object pose estimators for RGB images. To achieve this goal, we have designed:
   1. A novel correspondence-based framework using progressive dense keypoint localization and graph neural networks
   2. A novel visibility-aware importance measurement for object keypoints

Our methods have demonstrated the importance of deep geometric understanding for instance-level object pose estimation. Our next step is incorporate deep geometric understanding into novel object pose estimation.

2. Object pose estimation from cryo-EM images: The goal of this project is to develop efficient pose estimation methods for cryo-EM images and facilitate the 3D reconstruction process. To achieve this goal, we have designed an end-to-end learnable pipeline for orientation estimation. Our next step is to develop efficient algorithms for full pose estimation (i.e. orientation + translation) and evaluate on the CESPED dataset.
3. 2D/3D segmentation of optical coherence tomography: The goal of this project is to develop efficient segmentation methods to facilitate downstream OCT tasks.

Principal Investigator

Ruyi Lian, assistant professor, McComish Department of Electrical Engineering and Computer Science, South à£à£Ö±²¥Ðã State University

Publications

View Dr. Lian’s publications on .

Open positions:

The 3D Perception and Scientific Imaging Laboratory has open positions for Ph.D. candidates in 3D computer vision tasks and interdisciplinary topics. The preferred experiences and skills include:

1. Strong programming skills in Python and C++
2. Experience with deep learning frameworks (e.g., PyTorch)
3. Background in computer vision, 3D geometry or related areas
4. Familiarity with scientific imaging modalities (e.g., cryo-EM, OCT) is a plus
5. Good communication and teamwork abilities

If you are interested, please email Ruyi Lian with your CV and transcripts.