Zhixin Sun

Nanjing University of Posts and Telecommunications

Papers

Total Citations

H-Index

About

Zhixin Sun is a leading researcher in 3D computer vision and deep learning, with a particular focus on multi-view shape analysis and recognition. Their most influential work, "VERAM: View-Enhanced Recurrent Attention Model for 3D Shape Classification" (2018, 81 citations), addresses a critical limitation in multi-view neural networks: the lack of intelligent view selection. While existing approaches rely on simple max or average pooling to fuse features from different angles, Sun introduced a recurrent attention mechanism that dynamically selects and weighs the most informative views for 3D shape classification. This innovation not only improved classification accuracy but also enabled practical applications like active object recognition in robotics, where a system must decide which views to capture next. Sun's contributions have been widely recognized, with their work cited across computer vision, robotics, and graphics communities. By bridging the gap between passive multi-view learning and active perception, Sun has opened new avenues for research in 3D understanding, demonstrating how attention mechanisms can enhance both performance and real-world applicability in shape analysis tasks.