1. LCS: Learning Compressible Subspaces for Adaptive Network Compression at Inference Time Elvis Nunez, Maxwell Horton, Anish Prabhu, Anurag Ranjan, Ali Farhadi, and Mohammad, Rastegari arXiv preprint arXiv:2110.04252 2021 [arXiv]

    LCS is a method for training a "compressible subspace" of neural networks that contains a fine-grained spectrum of models that range from highly efficient to highly accurate. These models require no retraining, thus the subspace of models can be deployed entirely on-device to allow adaptive network compression at inference time.

  2. Token Pooling in Vision Transformers Dmitrii Marin, Jen-Hao Rick Chang, Anurag Ranjan, Anish Prabhu, Mohammad Rastegari, and Oncel, Tuzel arXiv preprint arXiv:2110.03860 2021 [arXiv]

    Token Pooling is a novel downsampling operator for vision transformers that efficiently exploits redundancies in the images and intermediate token representations. Applied to DeiT, it achieves the same ImageNet top-1 accuracy using 42% fewer computations.

  3. Hypersim: A photorealistic synthetic dataset for holistic indoor scene understanding Mike Roberts, Jason Ramapuram, Anurag Ranjan, Atulit Kumar, Miguel Angel Bautista, Nathan Paczan, Russ Webb, and Joshua M, Susskind In Proceedings of the IEEE/CVF International Conference on Computer Vision 2021 [arXiv] [Code] [URL]

    Hypersim is a photorealistic synthetic dataset for holistic indoor scene understanding containing 77,400 images of 461 indoor scenes with detailed per-pixel labels and corresponding ground truth geometry.

  1. MorphGAN: One-Shot Face Synthesis GAN for Detecting Recognition Bias Nataniel Ruiz, Barry-John Theobald, Anurag Ranjan, Ahmed Hussein Abdelaziz, and Nicholas, Apostoloff In arXiv 2020 [arXiv] [URL]

    MorphGAN can animate any face and control using a 3D rig. It’s one-shot and generalizes to in-the-wild unseen faces.

  2. GIF: Generative Interpretable Faces Partha Ghosh, Pravir Singh Gupta, Roy Uziel, Anurag Ranjan, Michael J. Black, and Timo Bolkart In International Conference on 3D Vision (3DV) 2020 [arXiv] [Code] [URL]

    GIF generates realistic face images and animate them with a 3D face rig.

  3. Learning Multi-Human Optical Flow Anurag Ranjan, David T Hoffmann, Dimitrios Tzionas, Siyu Tang, Javier Romero, and Michael J Black International Journal of Computer Vision 2020 [arXiv] [URL]
  4. Learning to Dress 3D People in Generative Clothing Qianli Ma, Jinlong Yang, Anurag Ranjan, Sergi Pujades, Gerard Pons-Moll, Siyu Tang, and Michael J. Black In Computer Vision and Pattern Recognition (CVPR) 2020 [arXiv] [Code] [URL]

    CAPE is a Graph-CNN based generative model for dressing 3D meshes of human body. It is compatible with the popular body model, SMPL, and can generalize to diverse body shapes and body poses. It is designed to be "plug-and-play" for many applications that already use SMPL. The CAPE Dataset provides SMPL mesh registration of 4D scans of people in clothing, along with registered scans of the ground truth body shapes under clothing.

  1. Towards Geometric Understanding of Motion

    Thesis. University of Tuebingen Tuebingen. 2019. [URL]
  2. Attacking Optical Flow Anurag Ranjan, Joel Janai, Andreas Geiger, and Michael J Black In International Conference on Computer Vision (ICCV) 2019 [arXiv] [Code] [URL]

    Deep learning based optical flow methods are vulnerable to adversarial attacks. We show that it is very easy to attack these systems in real world by just placing a small printed patch in the scene.

  3. Competitive Collaboration: Joint Unsupervised Learning of Depth, Camera Motion, Optical Flow and Motion Segmentation Anurag Ranjan, Varun Jampani, Lukas Balles, Kihwan Kim, Deqing Sun, Jonas Wulff, and Michael J Black In IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2019 [arXiv] [Code] [Talk]

    Unsupervised learning of several interconnected problems in low-level vision: single view depth prediction, camera motion estimation, optical flow and segmentation of a video into the static scene and moving regions.

  4. Unsupervised video segmentation Benjamin Ray, and Anurag Ranjan 2019 [URL]
  5. Capture, Learning, and Synthesis of 3D Speaking Styles Daniel Cudeiro, Timo Bolkart, Cassidy Laidlaw, Anurag Ranjan, and Michael J Black In IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2019 [arXiv] [Code] [URL]

    A neural network for generating 3D facial motion by using raw speech audio. Works on a veriety of unseen faces.

  1. Generating 3D faces using convolutional mesh autoencoders Anurag Ranjan, Timo Bolkart, Soubhik Sanyal, and Michael J Black In European Conference on Computer Vision (ECCV) 2018 [arXiv] [Code] [URL]

    A non-linear model for generating 3D faces using a Convolutional Autoencoder that operates directly on meshes. Our model is state of the art in generating diverse range of 3D facial meshes.

  2. Learning human optical flow Anurag Ranjan, Javier Romero, and Michael J Black British Machine Vision Conference (BMVC) 2018 [arXiv] [Code] [URL]

    Learning optical flow for humans is difficult. So, we created a synthetic dataset with realistic humans and trained a neural network on it.

  3. Unsupervised learning of multi-frame optical flow with occlusions Joel Janai, Fatma Guney, Anurag Ranjan, Michael Black, and Andreas Geiger In European Conference on Computer Vision (ECCV) 2018 [PDF] [Code]

    We propose a framework for unsupervised learning of optical flow and occlusions over multiple frames. Our multi-frame, occlusion-sensitive formulation outperforms existing unsupervised two-frame methods and even produces results on par with some fully supervised methods.

  1. Optical flow estimation using a spatial pyramid network Anurag Ranjan, and Michael J Black In IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2017 [arXiv] [Code] [URL]
  2. Seeing Skin in Reduced Coordinates Debanga R Neog, Anurag Ranjan, and Dinesh K Pai In IEEE International Conference on Automatic Face & Gesture Recognition (FG) 2017 [URL]

  1. Interactive gaze driven animation of the eye region Debanga R Neog, João L Cardoso, Anurag Ranjan, and Dinesh K Pai In International Conference on Web3D Technology 2016 [URL]

  1. Learning periorbital soft tissue motion

    Thesis. University of British Columbia,. 2015. [URL]

    We model the soft tissues around the eyes that are associated with subtle and fast motions and convey emotions during facial expressions. Our data driven model that can efficiently learn and reproduce the complex motion of these periorbital soft tissues.

  2. Gaze driven animation of eyes Debanga Raj Neog, Anurag Ranjan, João L Cardoso, and Dinesh K Pai In ACM SIGGRAPH/Eurographics Symposium on Computer Animation 2015 [PDF]

  1. Implementation of 3D object recognition and tracking Pankaj Bongale, Anurag Ranjan, and Sahil, Anand In International conference on Recent Advances in Computing and Software Systems (RACSS) 2012 [PDF]