Kaidi Cao

I am a senior undergraduate at Department of Electronic Engineering, School of Information Science and Technology, Tsinghua University.

During my internship at Sensetime Group Ltd. since Jul. 2016, I conducted research on the frontier of computer vision and deployed industrial applications with Cheng Li (Sensetime) and Prof. Chen Change Loy (CUHK). In Summer 2017, I spent 3 months at Stanford Concurrent VLSI Architecture Group as a Undergraduate Visiting Research Intern under the supervision of Prof. Bill Dally. Since Sept. 2017, I worked as a remote research assistant in Prof.Horowitz's VLSI Research Group. Currently I am a research intern at Visual Computing Group of Microsoft Research Lab - Asia, advised by Dr. Jing Liao and Dr. Lu Yuan.

My research interests lie primarily in the area of Computer Vision and Machine Learning.

Email / CV / LinkedIn / Github

Experience
tsinghua tsinghua tsinghua tsinghua
Education
tsinghua

Aug. 2014 - Jul. 2018 (Expected), Department of Electronic Engineering, Tsinghua University ,

Balchlor of Engineering, Electronic Information.

stanford

Jun. 2017 - Sep. 2017, Department of Computer Science, Stanford University ,

International Undergraduate Visiting Research Intern(UGVRI).

Selected Projects
halide_hls Halide-HLS with double buffer
Xuan Yang, Kaidi Cao, Mark Horowitz
Sep. 2017 - Present, Stanford University
[code]

Perfected lowering passes, wrote examples and testbenches for this system, ran tests to guarantee functionality doesn't Change, fixed multiple bugs.

pruning Efficient Channel Pruning
Kaidi Cao, Song Han, William J. Dally
Jun. 2017 - Sept. 2017, Stanford University

Proposed a pruning pipeline with filter selection and caliberation in a bottom up, layer by layer sequence. This work is deployed in production.

detection Efficient Video Detection
Kaidi Cao, Song Han, William J. Dally
Jun. 2017 - Sept. 2017, Stanford University

Proposed a pipeline to aggregate deep feature from key frame and staged shallow feature selected by comparing summary map from other frames to speed up detection.

Publications
cluster_pipeline

Merge or Not? Learning to Group Faces via Imitation Learning
Yue He, Kaidi Cao, Cheng Li, Chen Change Loy
AAAI Conference on Artificial Intelligence (AAAI), 2018
[pdf]

We proposed a novel face grouping framework that makes sequential merging decision based on short- and long-term rewards via inverse reinforcement learning.

Honors and Awards
  • Academic Excellent Scholarship, Tsinghua, Univ, 2017

  • Academic Excellent Scholarship, Tsinghua, Univ, 2016


This guy designs a nice homepage.