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
Top Papers
- 1
- 2A Generative Approach Towards Improved Robotic Detection of Marine Litter25 citations · 2020
- 3Person-following by autonomous robots: A categorical overview13 citations · 2019
- 4TrashCan 1.0 An Instance-Segmentation Labeled Dataset of Trash Observations12 citations · 2019
- 5ROW-SLAM: Under-Canopy Cornfield Semantic SLAM11 citations · 2022
- 6Trash-ICRA19: A bounding box labeled dataset of underwater trash11 citations · 2018
- 7
- 8
- 9Robotic Detection of Marine Litter Using Deep Visual Detection Models7 citations · 2019
- 10