Hongtao Wen (文洪涛) PhD Student Department of Computing, The Hong Kong Polytechnic University Location: P505, Mong Man Wai Building, The Hong Kong Polytechnic University Research Interest | Education and Intern | Publications | Projects | Skills | Honors & Awards | Misc Email: ht.wen@outlook.com (primary)        hongtao.wen@connect.polyu.hk [CV_HWen.pdf] [GitHub] [Google Scholar] [CSDN] [Wechat]

Research Interest

I work in the field of Robotics, Computer Vision and Deep Learning.
Currently, I focus on the following research topics:

• Robotic Perception
• Category-Level 6D Object Pose Estimation
• 6DoF Grasp Pose Estimation

Education and Intern

• 2023.9-NOW        PhD Student in COMP, The Hong Kong Polytechnic University.
• 2023.1-2023.8      Research Assistant in DeepSE Lab at The Hong Kong University of Science and Technology (HKUST).
• 2022.3-2022.6      AI Research Intern in Gongyuan Sanqian Technology Co., Ltd.
• 2020.9-2023.6      M.E. in SICE, Dalian University of Technology. Academic GPA 87.6/100
• 2016.9-2020.6      B.E. in SICE, Dalian University of Technology. GPA 87.3/100, top 10%

Publications

Conferences:

TransGrasp: Grasp Pose Estimation of a Category of Objects by Transferring Grasps from Only One Labeled Instance Hongtao Wen*, Jianhang Yan*, Wanli Peng, Yi Sun European Conference on Computer Vision 2022 (ECCV 2022), * indicates equal contribution. Webpage / Introduction Video / Code / PDF We propose TransGrasp, a category-level grasp pose estimation method that predicts grasp poses of a category of objects by labeling only one object instance.

Self-Supervised Category-Level 6D Object Pose Estimation with Deep Implicit Shape Representation Wanli Peng, Jianhang Yan, Hongtao Wen, Yi Sun Thirty-sixth AAAI Conference on Artificial Intelligence (AAAI 2022) Webpage / Introduction Video / Code / PDF We present a self-supervised method for category-level 6D pose estimation, SSC-6D, which can predict unseen object poses without explicit pose annotations and exact 3D models in real scenarios for training.

Patents:

• Updaing ...

Projects

Research Projects:

     These videos demonstrate some specific cases where TransGrasp predicts robust grasp poses for robotic manipulation.

Robot Demonstration: robot-assisted watering mobility-impaired individuals.	Robot Demonstration: autonomously pouring water from cup into bowl.	Robot Demonstration: autonomously grasping household objects.

Open-Source Projects:

• Vision-Based Robotic Grasping: A continuously updated summary of papers related to vision-based grasping, where I help to write a convenient program to automatically download these papers without human interference.
• Paddle ViT Tutorial: The repository provides demonstration codes for researchers to implement Vision Transformer(ViT) using an industrial deep learning framework called PaddlePaddle step by step.
• Paddle PiT: An unofficial repository implemented by PaddlePaddle of Spatial Dimensions of Vision Transformers, which is a Pooling-based Vision Transformer (PiT).
• Paddle GreedyHash: An unofficial repository implemented by PaddlePaddle of Greedy Hash: Towards Fast Optimization for Accurate Hash Coding in CNN, which aims to tackle the NP hard problem in Deep Hashing.
• Paddle HashNet: An unofficial repository implemented by PaddlePaddle of HashNet: Deep Learning to Hash by Continuation, which is a novel deep architecture for deep learning to hash by continuation method with convergence guarantees.
• More open-source contents can be found on GitHub and 百度 AI Studio.

Skills

• Excellent in project analysis, implementation and coordination.
• Language: Chinese (native proficiency) & English (proficiency, IELTS score 7.0).
• Programming language: Python, C++, C.
• In-depth knowledge of Robotics, Computer Vision and Deep Learning.
• Familiar with deep learning frameworks: PyTorch, PaddlePaddle.
• Familiar with the programming of robot ROS system.
• Self-motivated with passion for researches and technologies.

Honors & Awards

• 2022, Outstanding Graduate Student of DUT.
• 2021&2022, First Prize Scholarship of DUT.
• 2022, Champion on 3 tracks of Research Re-implementation Challenge by PaddlePaddle.
• 2021, Top 1% on 2 tracks of Challenge of Xunfei AI Recognition Algorithm.
• 2021, Top 1% on 3 tracks of Challenge of Guangdong Intelligent Recognition Algorithm.
• 2020, Top 1.5% on Guangdong Industrial Intelligent Manufacturing Innovation Contest -Intelligent Algorithm Tournament.
• 2020, Outstanding Graduate of Liaoning Province.
• 2020, Outstanding Graduate of DUT.
• 2019, China Aerospace Science and Technology Corporation Scholarship.
• 2018, Huawei Scholarship.
• 2017, First Prize Scholarship of Lingshui.
• 2017, Second Prize in Mathematics Competition of Dalian.
• 2017&2018&2019, Outstanding Merit Student of DUT.

Misc

• Swimming (breaststroke). I won the 4th place in the final of DUT swimming competition.
• Mountain climbing. Mountains and hills are everywhere in Dalian.
• Folk dance. I have participated in DUT art performance (Fenglan Cup) for 3 times.