Xingchao Peng

Papers

Total Citations

242

H-Index

About

Xingchao Peng is a leading researcher in computer vision and robotics, specializing in domain adaptation and transfer learning. His work addresses the critical challenge of bridging the gap between synthetic and real-world data, enabling machine learning models to generalize across different visual environments. Peng is best known for creating the VisDA benchmark (2018), a seminal synthetic-to-real dataset for visual domain adaptation that has garnered over 180 citations and become a standard evaluation tool in the field. His research has significantly advanced the practical deployment of AI in robotics, particularly through his work on adapting deep visuomotor representations from simulated to real environments (2015, 61+ citations). By developing methods that allow robots to transfer skills learned in simulation to physical world tasks—without costly real-world data collection—Peng has made foundational contributions to reducing the expense and risk of robotic training. His work continues to shape how researchers approach domain shift problems, with lasting impact on autonomous systems, visual recognition, and real-world AI deployment.