Chengxi Ye

University of Maryland, College Park

Papers

9

Total Citations

235

H-Index

6

About

Chengxi Ye is a computer vision and robotics researcher whose work spans event-based sensing, deep learning, and functional scene understanding. Best known for pioneering contributions to event camera perception, Ye led the development of EV-IMO — the first event-based dataset and learning pipeline for motion segmentation in indoor scenes — which has garnered nearly 100 citations and remains a landmark resource in the field. His research on unsupervised learning of dense optical flow, depth, and egomotion from sparse event data, published across multiple venues with a combined 60+ citations, demonstrated that lightweight neural architectures like the encoder-decoder network ECN could extract rich scene geometry from the unconventional output of Dynamic Vision Sensors. Beyond event cameras, Ye has contributed to cognitive robotics through his work on functional scene understanding, enabling robots to reason about actionable affordances rather than simple object labels. He also developed LightNet, a user-friendly, Matlab-based deep learning framework designed to lower barriers to entry for deep learning research. Collectively, Ye's portfolio reflects a commitment to building practical, deployable perception systems for autonomous agents navigating complex, dynamic environments.

Research Focus

Key Achievements

6
H-Index
9
Papers
235
Total Citations
26
Avg Citations/Paper
🏆 Most Cited Paper
EV-IMO: Motion Segmentation Dataset and Learning Pipeline for Event Cameras
97 citations · 2019
📈 Most Prolific Year: 2019 (3 Papers)
🤝 Key Collaborators: 9
🏛 Institutions: University of Maryland, College Park

Top Papers

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
    LightNet
    15 citations · 2016
  6. 6
  7. 7
    EV-IMO: Motion Segmentation Dataset and Learning Pipeline for Event Cameras
    6 citations · 2019
  8. 8
  9. 9

Key Collaborators

Contact & Links

Available for collaboration
Content generated · 14 days ago