Songle Chen

Nanjing University of Posts and Telecommunications

Papers

Total Citations

H-Index

About

Songle Chen is a leading researcher in 3D computer vision and deep learning, with a particular focus on multi-view shape analysis and intelligent perception. His most influential work, "VERAM: View-Enhanced Recurrent Attention Model for 3D Shape Classification" (2018, 81 citations), tackles a critical limitation in multi-view neural networks: the lack of a view selection mechanism. While standard approaches rely on simple max or average pooling to fuse features from multiple viewpoints, Chen’s VERAM introduces a recurrent attention model that dynamically selects and enhances informative views, dramatically improving classification accuracy and enabling active object recognition for robotic systems. This innovation bridges the gap between static 3D shape understanding and real-world, interactive perception. Beyond this flagship paper, Chen’s broader contributions to 3D deep learning have earned him over 1,500 total citations, reflecting the practical impact of his work in fields ranging from autonomous navigation to augmented reality. His research continues to push the boundaries of how machines see and interpret three-dimensional environments, making him a key figure in the evolution of intelligent visual systems.