Mohammad Shafiei Rezvani Nezhad

adjkd7@r.postjobfree.com +1-858-***-**** https://www.linkedin.com/in/mohammadshafiei/ Summary

More than 5 years of research experience in Computer Vision and 2 years in Machine Learning. I have lead two successful projects that lead to publications with publicly available results superior that the state of the art. Education

PhD Coursework, Computer Science & Engineering, UC San Diego September 2018-Present M.S., Visual Computing, University of Saarland (Germany) 2017 B.S., [subject/major], Qazvin I. Azad University (Iran) 2014 Skills

Programming languages: C++, Python, Matlab.

Development environments: VS Code.

Machine learning tools: Pytorch, OpenCV, Numpy,

Tensorboard, WandB, Kubernetes, Ceres.

Computer Vision libraries: Point Cloud Library, CUDA.

Computer graphics libraries: Mitsuba, VTK, CGAL.

Theoretical skills: Linear algebra,

Optimization. Machine Learning, Computer

Vision, Image Processing

Languages: English (Fluent), German

(intermediate), Persian (native speaker)

Experience

Research Assistant at Visual Computing Lab., University of California San Diego, USA Sep. 2018 – present

● Lead ongoing research project between research institutes on Specular Scene Reconstruction and Relighting with superior than state of the art results.

● Implemented and evaluated rendering layer in a neural network architecture designed for inverse rendering with code and documentation. Rendering results are compared and shown to be of higher quality than the state of the art.

● Collaborated with team of researchers to write and publish paper on inverse rendering for complex indoor scenes.

Computer Vision Research Intern, Fyusion Inc., San Francisco, USA Jun – Sep 2020

Initiated and led a project on Image Based Relighting of highly specular objects.

Used Python, Pytorch, OpenCV, numpy, and other libraries and GPU clusters to train machine learning methods.

Evaluated the baseline and state of the art methods Graduate Research Scientist Intern, Adobe Inc., San Jose, USA Jun – Sep 2019

Initiated and led a project on Lighting Design: I created an optimization-based system enabling artists to design complicated 3D lighting setups for 3D models using simple 2D annotations.

Used Mitusba, OpenCV, Ceres solvers.

Research Assistant Student at Max Planck Institute of Informatics, Saarland, Germany.

Jan. 2016 – Jan. 2018

Initiated and led project leading to publication on the topic of live illumination decomposition of videos.

● Implemented inverse kinematic in C++ for single image motion capture project; collaborated with research team to publish paper on 3D human pose estimation

● Implemented camera trajectory estimation for human motion capture using OpenMVG library project; published paper on results.

Off-shore Software Engineer at Photocore AG, Zürich, Switzerland. Nov. 2014 – May 2015

● Implemented visualization and interaction tools for point clouds using PCL and VTK libraries. Research Assistant Student at Qazvin Azad University, Qazvin, Iran Sep 2008 - Sep 2014

● Initiated and lead a project published code and documentation in “Ground-truth localization system for humanoid soccer robot using RGB-D camera”, Mohammad Shafiei Rezvani Nezhad, Omid Amirghiasvand, International Journal of Computational Vision and Robotics, Volume 7 Issue 3, January 2017, Pages 285-301, code and the paper accessible at https://github.com/mrlspl/GroundTruth.

● Calibrated and acquired RGB-D images from Asus Xtion cameras

● Competed in Standard Platform League (SPL) Robocup international and local competitions (Singapore, Turkey, Mexico and Brazil).

Awards and Selected Publications

Awards

Scholarship, Azad University, Qazvin Branch, 2008-2014. 3rdplace (out of 10 international teams), Standard Platform League, German Open competitions, Magdeburg, Germany, April 2014.

1st place (out of 10 international teams), Best Research Award in Standard Platform League, Iran Open competitions 2013, Tehran, Iran.

Publications

● “Live Illumination Decomposition of Videos”, Abhimitra Meka*, Mohammad Shafiei*, Michael Zollhöfer, Christian Theobalt, arXiv 2019, https://arxiv.org/pdf/1908.01961.pdf (* equal contribution, accepted to ACM TOG Journal)

● “VNect: Real-time 3D Human Pose Estimation with a Single RGB Camera”, Dushyant Mehta, Srinath Sridhar, Oleksandr Sotnychenko, Helge Rhodin, Mohammad Shafiei, Hans-Pieter Seidel, Weipeng Xu, Dan Casas, Christian Theobalt, SIGGRAPH 2017.

● “EgoCap: Egocentric Marker-less Motion Capture with Two Fisheye Cameras”, H. Rhodin, C. Richardt, D. Casas, E. Insafutdinov, Mohammad Shafiei, Hans-Peter Seidel, Bernt Schiele, Christian Theobalt, SIGGRAPH Asia 2016.

Contact this candidate