Papers
10
Total Citations
624
H-Index
8
About
Thanh-Toan Do is a prominent researcher in computer vision and robotics, with a particular focus on object detection, pose estimation, and robotic perception. His work spans deep learning-driven approaches to enable robots to better understand and interact with their physical environments. Do's most influential contribution is AffordanceNet (2018), an end-to-end deep learning framework that simultaneously detects objects and their affordances from RGB images, amassing over 290 citations and establishing a foundational benchmark in affordance detection. His Deep-6DPose work (2018, 129 citations) further advanced the field by recovering full 6D object poses from single RGB images — a critical capability for robotic manipulation. A defining theme of Do's more recent research is transparent object perception, an exceptionally challenging problem due to light reflection and refraction. His comprehensive review on the topic (2023, 55 citations) alongside innovations like A4T and vision-guided tactile poking approaches demonstrate his commitment to solving real-world robotic handling challenges through creative sensor fusion and hierarchical learning strategies. With his RoTipBot system and V2CNet framework for video-to-command translation, Do consistently bridges the gap between visual perception and actionable robotic control, making his body of work essential reading for researchers in intelligent robotics and embodied AI.
Research Focus
Key Achievements
Top Papers
- 1
- 2Deep-6DPose: Recovering 6D Object Pose from a Single RGB Image129 citations · 2018
- 3Robotic Perception of Transparent Objects: A Review55 citations · 2023
- 4
- 5
- 6
- 7
- 8
- 9SceneCut: Joint Geometric and Object Segmentation for Indoor Scenes6 citations · 2018
- 10