Angjoo Kanazawa

I am a BAIR postdoc at UC Berkeley adviced by Jitendra Malik, Alyosha Efros and Trevor Darrell. I did my PhD in CS at the University of Maryland, College Park with my advisor David Jacobs. Prior to UMD, I spent four wonderful years at NYU where I had the chance to work with Rob Fergus and completed my BA in Mathematics and Computer Science. Since then I've had the pleasure to work with: Peter Belhumeur at Columbia University Google[x] self-driving car team Manmohan Chandraker at NEC Labs in Cupertino Michael Black at Max Planck Institute in Tübingen, Germany Resume/CV

Contact

Email: kanazawa at eecs.berkeley.edu
Homepage: this page | old

Research Interest

My work focuses on 3D reconstruction of animals including humans from single-view images. I work with animals because I am interested in how to represent non-rigid 3D shapes that deform and articulate. In particular I'm interested in how we can build a model of non-rigid 3D shapes from 2D images or videos. Also, animal pictures are a lot of fun to work with.

I believe the key to solving this problem is having a strong prior on object pose and shape, which can resolve many of the ambiguities. I am also interested in exploring how deep learning methods can be used for building such models.

Papers

• 3D Menagerie: Modeling the 3D shape and pose of animals
  Silvia Zuffi, Angjoo Kanazawa, David Jacobs, Michael J. Black
  Computer Vision and Pattern Recognition (CVPR) 2017. (Spotlight)
  [pdf] [arXiv]

• Keep it SMPL: Automatic Estimation of 3D Human Pose and Shape from a Single Image
  Federica Bogo*, Angjoo Kanazawa*, Christoph Lassner, Peter Gehler, Javier Romero and Michael J. Black
  (* equal contribution)
  European Conference on Computer Vision (ECCV) 2016. (Spotlight)
  [pdf] [project page with code] [Spotlight video ]

  • Our work was featured in two minute papers

• WarpNet: Weakly Supervised Matching for Single-View Reconstruction
  Angjoo Kanazawa, David W. Jacobs, Manmohan Chandraker
  Computer Vision and Pattern Recognition (CVPR) 2016.
  [pdf] [supp] [test set ids & our curves]

• Learning 3D Deformation of Animals from 2D Images
  Angjoo Kanazawa, Shahar Kovalsky, Ronen Basri, David W. Jacobs
  Eurographics 2016. Günter Enderle Best Paper Award
  [pdf] [code on github] [fastforward] [cat results]

• Locally Scale-invariant Convolutional Neural Network
  Angjoo Kanazawa, Abhishek Sharma, David W. Jacobs
  Deep Learning and Representation Learning Workshop: NIPS 2014.
  [pdf][code on github]

• Affordance of Object Parts from Geometric Features
  Austin Myers, Angjoo Kanazawa, Cornelia Fermuller, Yiannis Aloimonos
  RGB-D: Advanced Reasoning with Depth Cameras: RSS 2014 [pdf]
  Vision Meets Cognition Workshop: CVPR 2014 [pdf]

• Dog Breed Classification Using Part Localization
  Jiongxin Liu, Angjoo Kanazawa, Peter Belhumeur, David W. Jacobs
  European Conference on Computer Vision (ECCV), Oct. 2012.
  [pdf] [slides]
  

  • try our iPhone app: Dogsnap !

  • Columbia dogs with parts dataset used in the paper: zip file (1.1GB)

    • 133 breeds recognized by the American Kennel Club

    • 8,351 images of dogs from Google image search, Image-net, and Flickr.

    • 8 part locations annotated for each image

Thesis

Single-View 3D Reconstruction of Animals
Angjoo Kanazawa
Doctoral Thesis, University of Maryland, August 2017
[pdf]

Teaching

• Teaching Assistant: CMSC 421 Spring 2012 Introduction to Artificial Intelligence

• Teaching Assistant: CMSC 131 Fall 2011 Object-Oriented Programming I

• Teaching Assistant: CSCI-UA.0101,0103, Fall 2008, Spring 2009 Introduction to Computer Science I, II