catree / ResearchWork

My research work, mainly on 3D reconstruction, includes Structure from Motion, SLAM, semantic reconstruction

ResearchWork

My personal research work on Peking University, mainly on 3D reconstruction, includes Structure from Motion (SfM), Simultaneous Localization and Mapping (SLAM), semantic reconstruction. I'm also interested in numerical optimization, especially large scale nonlinear optimization, though only at the begin right now. My dream is to build the world with computer vision technologies. If you are interested in my work, just send me an e-mail.

1. Structure from Motion

Incremental SfM

• [3DV 2013] Towards Linear-time Incremental Structure from Motion

• [CVPR 2016] Structure-from-Motion Revisited

• [3DV 2017] Batched Incremental Structure-from-Motion

• [ECCV 2018] Progressive Structure from Motion

Global SfM

Hybrid SfM

• [ACCV 2014] Divide and Conquer: Efficient Large-Scale Structure from Motion Using Graph Partitioning

• [arXiv 2017] Parallel Structure from Motion from Local Increment to Global Averaging

• [CVPR 2018] Very Large-scale Global SfM by Distributed Motion Averaging

Hierarchical SfM

Viewing Graph

• [ICCV 2003] video google: a text retrieval approach to object matching in videos

• [CVPR 2006] Scalable Recognition with a Vocabulary Tree

• [CVPR 2008] Skeletal graphs for efficitent structure from motion

• [ICCV 2009] Build Rome in a Day

• [ECCV 2010] Improving the fisher kernel for large-scale image classification

• [IJCV 2013] Image Classification with the Fisher vector: theory and practice

• [ICCV 2015] Optimizing the Viewing Graph for Structure-from-Motion Brief Comment: This approach has a strong geometric verification by enforcing loop consistency, while not efficiency and may more suitable for global SfM other than other SfM approaches. Besides, the implementation is trivially.

• [ECCV 2016] Graph-Based Consistent Matching for Structure-from-Motion

• [3DV 2017] GraphMatch: Efficient Large-Scale Graph Construction for Structure from Motion

• [3DV 2018] Progressive Large-Scale Structure-from-Motion with Orthogonal MSTs

• [ECCV 2018] View-graph Selection Framework for SfM

Bundle Adjustment

TODO:

• [CVPR 2016] A Consensus-Based Framework for Distributed Bundle Adjustment

• [arxiv 2017] Distributed Bundle Adjustment

• [ICCV 2017] Distributed Very Large Scale Bundle Adjustment by Global Camera Consensus

• [SIAM 2011] Distributed Optimization and Statistical Learning via the Alternating Direction Method of Multipliers

2. Simultaneous Localization and Mapping

3. Semantic Reconstruction

4. My Pulications

• [PRCV 2018] Yu Chen, Yao Wang, Youkun Wu, Peng Lu, Yisong Chen and Guoping Wang. Large-Scale Structure from Motion with Semantic Constraints of Aerial Images

5. My Open-source Code

• GraphSfM

• Graph Cluster

• ComputerVisionDatasets

• MyCodeSnippet

• Code Practice