Zhengyin Du

Papers

Total Citations

H-Index

About

Zhengyin Du is a researcher advancing the field of video understanding, with a primary focus on temporal action segmentation and fine-grained human activity analysis. Their most notable contribution is the development of the Atrous Temporal Convolutional Network (ATCN), introduced in a 2019 paper that has garnered 13 citations. This work addresses a critical challenge in untrimmed video analysis: the need for robust action segmentation across varying temporal scales. By leveraging atrous convolutions, Du’s architecture effectively captures multi-scale temporal dependencies, enabling precise segmentation of actions in surveillance, robotics, and other real-world applications. This innovation enhances the ability to parse continuous video streams into meaningful action units, a key step toward autonomous systems that understand human behavior. Du’s research sits at the intersection of computer vision and temporal modeling, offering practical solutions for video-based monitoring and human-robot interaction. Their work continues to influence the development of more resilient and accurate video analysis systems, making a tangible impact on how machines interpret dynamic human actions in unconstrained environments.