Zhengfei Song

Hi there! I’m Zhengfei Song, a first-year master’s student in Electrical Engineering at the California Institute of Technology (Caltech), within the Division of Engineering and Applied Science. I am honored to be supported by the Chiang Chen Overseas Fellowship ($50,000).

I received my Bachelor of Engineering in Automation from the College of Electronics and Information Engineering at Tongji University in June 2026. During my undergraduate studies, I conducted research in semantic segmentation with the Machine Intelligence & Autonomous Systems (MIAS) Group under the supervision of Prof. Rui Fan.

My research interests lie in intelligent systems, with a focus on perception, multimodal learning, and real-world applications of machine learning. My goal is to develop systems that integrate diverse sources of information to perceive and understand the world, enabling reliable and adaptive behavior in complex environments.

Zhengfei Song at Caltech

News

9/28📚 I begin my graduate studies at Caltech! Look forward to immersing myself in this vibrant research environment and pushing the limits of both myself and the systems and models I build.
8/15🔥 Our paper, “Vehicular road crack detection with deep learning: An online benchmark for comprehensive evaluation of existing algorithms,” has been accepted by Advanced Engineering Informatics (IF: 11.5)!

Selected Publications

Overview of the road crack detection benchmarkAdditional road crack detection benchmark figureVehicular road crack detection with deep learning: An online benchmark for comprehensive evaluation of existing algorithms
N. Ma, Z. Song, Q. Hu, C. W. Liu, Y. Han, Y. Zhang, R. Fan, L. Xie
Advanced Engineering Informatics 77, 105158, 2027 (IF: 11.5)
Paper  /  Website  /  Dataset
DualCrackNet overviewAdditional DualCrackNet figureRobust and Real-time Road Crack Detection through Collaborative Dual-Branch Learning on Robotic Sensing Platform
Z. Song, N. Ma, Z. Zhang, M. J. Lee, S. Vityazev, A. Dvorkovich, R. Fan
IEEE International Conference on Real-time Computing and Robotics, 2025
Paper  /  Code
LR-CrackNet overviewAdditional LR-CrackNet figureSelf-Derived Multi-Modal Knowledge Distillation for Real-time Road Crack Detection
N. Ma*, Q. Hu*, Z. Song*, S. Guo, R. Fan (*Equal Contribution)
IEEE Sensors Journal, 2025 (IF: 4.5)
Paper  /  Code

All publications

Visitors

Website template from Jon Barron.