Papers

13

Total Citations

165

H-Index

7

About

Jungseok Hong is a robotics and computer vision researcher whose work spans marine robotics, autonomous underwater vehicles (AUVs), and human-robot interaction. He is perhaps best known for developing the **TrashCan** dataset series, a pioneering contribution to underwater marine debris detection that provides thousands of semantically segmented and instance-labeled images for training robust visual detection systems. This work, which has accumulated over 57 citations for its 2020 iteration alone, has become a foundational resource for researchers tackling ocean pollution through autonomous robotics. Hong's research addresses critical data scarcity challenges in underwater vision by leveraging generative models, including variational autoencoders, to synthesize realistic training imagery. Beyond debris detection, his work extends to semantic SLAM for agricultural robotics, semantically-aware obstacle avoidance, and AUV-diver interaction using monocular vision and human body priors. His categorical overview of person-following robots further demonstrates his breadth across human-robot collaboration domains. A unifying thread throughout Hong's portfolio is the deployment of intelligent perception systems in difficult, real-world environments — whether beneath ocean surfaces or beneath corn canopies — making his contributions particularly valuable for researchers building practical, field-ready autonomous systems.

Research Focus

Key Achievements

7
H-Index
13
Papers
165
Total Citations
13
Avg Citations/Paper
🏆 Most Cited Paper
TrashCan: A Semantically-Segmented Dataset towards Visual Detection of Marine Debris
57 citations · 2020
📈 Most Prolific Year: 2020 (4 Papers)
🤝 Key Collaborators: 18
🏛 Institutions: University of Minnesota, University of Minnesota System, Twin Cities Orthopedics

Top Papers

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10

Key Collaborators

Contact & Links

Available for collaboration
Content generated · 15 days ago