Takayoshi Yamashita
Papers
17
Total Citations
181
H-Index
7
About
Takayoshi Yamashita is a computer vision and robotics researcher whose work sits at the intersection of deep learning, visual perception, and autonomous manipulation. His most significant contributions center on multi-task learning frameworks for robotic systems, most notably the MT-DSSD (Multi-Task Deconvolutional Single Shot Detector), which simultaneously performs object detection, semantic segmentation, and grasp-point estimation within a unified neural network — a landmark contribution to robot perception that has accumulated over 50 citations across its iterations. Yamashita has also made important strides in explainable AI, developing the Attention Branch Network, which leverages attention mechanisms not merely for visual explanation but to actively improve classification performance — a meaningful step beyond conventional interpretability methods. His research further explores deep reinforcement learning transparency, layered object detection for bin-picking scenarios, and collision-risk prediction for domestic service robots. Practical real-world robotics informs much of his agenda, including participation in Amazon Picking Challenge 2016 and automated data-collection pipelines using mobile robots. Collectively, his body of work demonstrates a sustained commitment to making robotic manipulation systems more perceptive, interpretable, and deployable in complex, unstructured environments.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4Detecting layered structures of partially occluded objects for bin picking21 citations · 2019
- 5
- 6
- 7
- 8
- 9Iterative Coarse-to-Fine 6D-Pose Estimation Using Back-propagation5 citations · 2021
- 10