Stephen David Miller

Startup cofounder, AI researcher, podcaster, person, etc.

Talks, Etc.

Last update:

Writing

  Description
“Forests, Trees, and False Dichotomies”
ACM XRDS Magazine, Summer 2017
[pdf]
———
Brief: Contributed feature article; misc advice on being a balanced developer, remaining open to new ideas

Videos

Most presentations seem to be recorded by the conference venue then vanish into the aether; but some have been preserved! Here are a few videos of talks I’ve given.

From “Research” to “Industry”: Mobile Edition Given at the Conference for Computer Vision and Pattern Recognition (CVPR) 2018 in Salt Lake City, UT. Should start at 29m04s TLDR; tips and tricks for scaling research code to production, woes of startup life, small bits of shameless self promotion

Fyusion: Immersive 3D Imaging at Scale Winner of the Plug n’ Play Spring 2018 Pitch Competition in Santa Clara, CA. Should start at 1h37m

Bigger Data vs Better Data: New Inputs, New Outputs, and How to Make Them Useful
Given at the FUEL by Medialaan event, 2017 in Brussels.

Clean Models From Noisy Sensors: RGBD Reconstruction
Given at the Stanford Berkeley Robotics Symposium (SBRS) 2013 in Berkeley, CA. Should start at 53m06s

Recent Advancements in PCL
Given at the Stanford Berkeley Robotics Symposium (SBRS) 2013 in Berkeley, CA. Should start at 35m08s

Perception for the Manipulation of Socks
Given at the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2011 in San Francisco, CA

Parametrized Shape Models for Clothing
Given at the IEEE International Conference on Robotics and Automation (ICRA) 2011 in Shanghai, China

Slides

Note: Powerpoint / Keynote slides tend to be massive, and fairly tough to port with videos. Here I’ll upload PDFs, but feel free to e-mail me (stephen__at__cs.stanford.edu) if you’d like to use slides in your presentation, and I’ll be happy to adapt to your software.

Preview Brief
CVPR 2018 From “Research” to “Industry”: Mobile Edition
30 minute talk about growing pains in transitioning from academia to startup life; goes over some software engineering tricks, architecture, management, process, git flow. Extremely light / high level.
PDF
Dense 2013 Dense reconstruction
Lecture for CS 287 at UC Berkeley. About 80 minutes; goes over the basic building blocks of Kinect Fusion, brief survey of subsequent research.
PDF
IROS 2013 Unsupervised extrinsic calibration of depth sensors in dynamic scenes
Given at the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2013 in Tokyo, Japan
12 minute talk on an unsupervised method for registering depth sensors together in time and space. Mostly high-level, lots of videos.
PDF
SBRS 2013 Clean Models from Noisy Sensors: RGBD Reconstruction
Given at the Stanford Berkeley Robotics Symposium (SBRS) 2013 in Berkeley, CA
5 minute spotlight for our ICCV work on elastic registration. No equations.
PDF
ICRA 2013 Clean models from noisy sensors: registration and reconstruction techniques
Given at the IEEE International Conference on Robotics and Automation (ICRA) 2013 PCL Tutorial in Karlsruhe, Germany
20 minute talk highlighting a few papers related to reconstruction, with an emphasis on code + PCL usage.
PDF
ICRA 2011 Parameterized Shape Models for Clothing
Given at the IEEE International Conference on Robotics and Automation (ICRA) 2011 in Shanghai, China
12 minute talk about the perception behind the Berkeley laundry folding robot
PDF