Congcong Li
Papers
1
Total Citations
45
H-Index
1
About
Congcong Li is a leading researcher in computer vision, with a primary focus on holistic scene understanding—the challenge of enabling machines to interpret all elements of a visual scene simultaneously. Her seminal work, "Towards Holistic Scene Understanding: Feedback Enabled Cascaded Classification Models" (2011, 45 citations), introduced a groundbreaking framework that integrates multiple interrelated sub-tasks—such as scene categorization, depth estimation, and object detection—into a single, feedback-driven system. By allowing classifiers to communicate and refine each other’s outputs, Li demonstrated that joint reasoning dramatically improves accuracy over isolated approaches. This work has been highly influential, laying the foundation for modern multi-task learning in vision and inspiring subsequent research on end-to-end scene parsing. Li’s contributions are especially notable for their practical impact: her cascaded models reduce computational redundancy while boosting performance, making them valuable for real-world applications like autonomous driving and augmented reality. With a citation count reflecting sustained relevance, Li continues to shape how researchers think about comprehensive visual perception, bridging the gap between individual vision tasks and truly intelligent scene interpretation.
Research Focus
Key Achievements
Top Papers
- 1