Shaofei Qin

Graduate · Information Engineering
800 Dongchuan RD. Minhang District, Shanghai, China
(+86) 189-1693-0119 | Sufer_Qin@163.com | 1105042987(Sufer_Qin)

Education

---

• B.S. in SJTU (Shanghai Jiao Tong University):
  Shanghai, China 2016.09 - 2020.06
  
  
  • Rank: 7/159;    Average Grades: 90.15;
• M.S. in SJTU (Shanghai Jiao Tong University):
  Shanghai, China 2020.09 - 2023.03

Skill

---

• Programming: Python, C/C++, CUDA, C#, Java, HTML5
• Languages: English (CET-6), Chinese (Native)

Research Experience

---

• RoboMaster:

  Embedded System Design:

  2017.06 - 2019.08
  
  
  • Designed a user-friendly program framework for the school team.
  • Researched for an new algorithm to optimize the stability of the gimbal of the robots.

• China Undergraduate IoT (Internet of Things) Design Competition:

  Dense QR code Decoder:

  2019.04 - 2019.08
  
  
  • Design a new net based on CenterNet which can increases operating speed by hundreds times with little loss of accuracy and enable the program to run in real time on mobile devices.
  • Design a new tracking algorithm that utilizes temporal information redundancy to reduce duplicate calculations and improve object detection accuracy.
  • Nearly 100 QR codes can be found and decoded on the mobile phone in nearly 10 FPS, which is suitable for environments where QR codes are densely present such as hospital.

• Undergraduate Graduation Design:

  Quaternion Product Units for Deep Learning on 3D Rotation Groups:

  2019.10 - 2020.06
  
  
  • Propose QPU for inferrence and optimization on 3D rotation group to make neural network get rid of the influence of input data rotation;
  • Based on the mathematical principle of Hamiltonian product, we designed an optimization strategy from the perspective of data structure and algorithm, and implemented it with CUDA, which increased the operating speed by 80%;
  • Designed a method for mapping from the quaternion field to the real field, which significantly improves the compatibility with the real network and facilitates the integration with existing research results.

• Recent Work:

  Skeleton Action Generation with Rotation Equivariant:

  2020.10 - Now
  
  
  • As a continuation of previous research ideas, we make the neural network output further have the ability to follow the input data to rotate synchronously;
  • Compared with the existing research, the proposed method can generate a sequence of actions in the same direction with the direction of the input data, which will reduce the dependence of network on the train data;
  • Further apply the above properties to action sequence interpolation, so that action sequences can be generated naturally from few key frames, thereby reducing the cost of motion capture when making 3D animations.

Extracurricular Activity

---

• Project Management Department, SIC(Student Innovation Center) of SJTU: Minister
  • Organized the RoboMaster School Tournament.
  • Provided technical support for embedded development courses at SIC.

Honors & Awards

---

• 2017.11 , National Scholarship , Chinese Ministry of Education.
• 2017.12 , First Prize in Regional College Students Physics Contest , Shanghai Physical Association.
• 2018.08 , Second Prize in RoboMaster 2018 National Finals , Central Committee of the Communist Youth League & National Federation of Students.
• 2018.11 , Academic Excellence Scholarship , Shanghai Jiao Tong University.
• 2018.12 , Excellent Student Cadre , Communist Youth League, Shanghai Jiaotong University Committee.
• 2019.08 , Second Place (Special prize) in RoboMaster 2019 National Finals , Central Committee of the Communist Youth League, National Federation of Students.
• 2019.08 , First Prize in China Undergraduate IoT (Internet of Things) Design Competition , Ministry of Education, Computer Science Teaching Steering Committee.
• 2019.08 , 2nd Place in Signet Ring Cell Detection task and 4th Place in the Colonoscopy Tissue Segmentation and Classification task at Digestive-System Pathological Detection and Segmentation Challenge 2019 , Grand Challenges & MICCAI 2019.
• 2019.11 , Academic Excellence Scholarship , Shanghai Jiao Tong University.
• 2020.06 , Outstanding graduates , Shanghai Jiao Tong University.

Publications

---

• Quaternion Product Units for Deep Learning on 3D Rotation Groups

  Xuan Zhang, Shaofei Qin, Yi Xu, Hongteng Xu

  IEEE/CVF Conference on Computer Vision and Pattern Recognition 2020 (CVPR 2020)

• Fast Batch Reading Densely Deployed QR Codes

  Xiaohua Tian*(Tutor), Shaofei Qin*, Binyao Jiang, Xinbing Wang

  IEEE Transactions on Mobile Computing 2021 (second review)

• Dominant Patterns: Critical Features Hidden in Deep Neural Networks

  Zhixing Ye*, Shaofei Qin*, Sizhe Chen, Xiaolin Huang

  Conference and Workshop on Neural Information Processing Systems 2021 (NIPS 2021, under review)