Yuzhuo Ren successfully passed her PhD Qualifying Exam on August 29, 2016. The title of her Ph.D. thesis proposal is “Outdoor and Indoor Layout Estimation Using Machine Learning Techniques”. Her Qualifying Exam committee consisted of Jay Kuo (Chair), Sandy Sawchuk, Antonio Ortega, Panos Georgiou and Aiichro Nakano (Outside Member).
In her proposal, two systems are proposed to address outdoor layout estimation problem and indoor layout estimation problem. A global-attributes assisted labeling (GAL) system is proposed to address outdoor layout estimation problem. The proposed GAL system exploits both local features and global attributes, which consists of three stages, namely, initial pixel labeling, global attributes extraction and label refinement. A coarse-to-fine indoor layout estimation (CFILE) method is proposed. The proposed CFILE method combines bottom-up knowledge from deep learning with top-down prior knowledge. The proposed GAL system and CFILE system achieve state-of-the-art performance.