Yuren Cong
Papers
2
Total Citations
5
H-Index
1
About
Yuren Cong is a rising researcher at the forefront of computer vision, with a focus on advancing scene understanding and segmentation in complex, real-world environments. Cong’s work bridges the gap between simulation and reality, particularly through the development of the Segment Any Object Model (SAOM). This innovative framework introduces a real-to-simulation fine-tuning strategy for the foundational Segment Anything Model (SAM), enabling robust multi-class multi-instance segmentation—a critical task for identifying and delineating multiple object classes and their individual instances within a single image. With 4 citations since 2024, SAOM addresses a key limitation of SAM, which often produces partial or incomplete masks, thereby pushing the boundaries of promptable segmentation. Cong also contributed to PanoSCU, a simulation-based dataset for panoramic indoor scene understanding. This dataset is pivotal for tasks like visual room rearrangement, where agents must restore objects to their original states, leveraging the comprehensive spatial awareness offered by 360-degree views. By tackling the challenges of panoramic perception and fine-grained segmentation, Yuren Cong is laying essential groundwork for more intelligent, context-aware robotic systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2