Joosoon Lee
Papers
2
Total Citations
69
H-Index
2
About
Joosoon Lee is a leading researcher in computer vision and robotics, specializing in amodal instance segmentation and object detection for unstructured environments. His most influential work, "Unseen Object Amodal Instance Segmentation via Hierarchical Occlusion Modeling" (2022), has garnered 66 citations and addresses a critical gap in robotic manipulation: the ability to segment not just visible regions but also occluded parts of unfamiliar objects. By introducing hierarchical occlusion modeling, Lee enables robots to perceive complete object shapes in cluttered scenes—a foundational capability for autonomous grasping and assembly tasks. This work stands out for its practical impact on real-world robotics, where objects are often partially hidden. Lee also contributed to "Object Detection for Understanding Assembly Instruction Using Context-aware Data Augmentation and Cascade Mask R-CNN" (2021), which advances robotic task planning by segmenting speech bubbles from 2D assembly diagrams. His research bridges perception and action, equipping robots with the visual understanding needed to operate in dynamic, unstructured settings. With a focus on unseen objects and occlusion reasoning, Lee’s work is shaping the next generation of intelligent robotic systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2