Jan Kautz is Vice President of Learning and Perception Research at NVIDIA. He and his team pursue fundamental research in the areas of computer vision and deep learning, including visual perception, geometric vision, generative models, and efficient deep learning. Their work has been given various awards and has been regularly featured in the media. Before joining NVIDIA in 2013, Jan was a tenured faculty member at University College London. He holds a degree in Computer Science from the University of Erlangen-Nürnberg (1999), an MMath from the University of Waterloo (1999), received his PhD from the Max-Planck-Institut für Informatik (2003), and worked as a post-doctoral researcher at the Massachusetts Institute of Technology (2003-2006). Jan has chaired numerous conferences (Eurographics Symposium on Rendering 2007, IEEE Symposium on Interactive Ray-Tracing 2008, Pacific Graphics 2011, CVMP 2012, Eurographics 2014) and has served on several editorial boards (ACM Transactions on Graphics, IEEE Transactions on Visualization & Computer Graphics, The Visual Computer, Computer Graphics Forum, International Journal of Image and Graphics). He is a member of ACM and IEEE.
Latest
- Compact Language Models via Pruning and Knowledge Distillation
- CosAE: Learnable Fourier Series for Image Restoration
- MaskLLM: Learnable Semi-Structured Sparsity for Large Language Models
- SpatialRGPT: Grounded Spatial Reasoning in Vision-Language Models
- COIN: Control-Inpainting Diffusion Prior for Human and Camera Motion Estimation
- DiffiT: Diffusion Vision Transformers for Image Generation
- LITA: Language Instructed Temporal-localization Assistant
- Flextron: Many-in-One Flexible Large Language Model
- AM-RADIO: Agglomerative Model - Reduce All Domains Into One
- COLMAP-Free 3D Gaussian Splatting
- FoundationPose: Unified 6D Pose Estimation and Tracking of Novel Objects
- GAvatar: Animatable 3D Gaussian Avatars with Implicit Mesh Learning
- Hydra-MDP: End-to-end Multimodal Planning with Multi-target Hydra-Distillation
- Is Ego Status All You Need for Open-Loop End-to-End Autonomous Driving?
- VILA: On pretraining for vision language models
- 3D Reconstruction with Generalizable Neural Fields using Scene Priors
- A Variational Perspective on Solving Inverse Problems with Diffusion Models
- FasterViT: Fast Vision Transformers with Hierarchical Attention
- Learning to Jointly Understand Visual and Tactile Signals
- FoVA-Depth: Field-of-View Agnostic Depth Estimation for Cross-Dataset Generalization
- PACE: Human and Camera Motion Estimation from in-the-wild Videos
- Convolutional State Space Models for Long-Range Spatiotemporal Modeling
- Generalizable One-shot Neural Head Avatar
- PhysDiff: Physics-Guided Human Motion Diffusion Model
- RANA: Relightable and Articulated Neural Avatars
- SMRD: SURE-based Robust MRI Reconstruction with Diffusion Models
- Single-Shot Implicit Morphable Faces with Consistent Texture Parameterization
- Global Context Vision Transformers
- Loss-Guided Diffusion Models for Plug-and-Play Controllable Generation
- BundleSDF: Neural 6-DoF Tracking and 3D Reconstruction of Unknown Objects
- FB-OCC: 3D Occupancy Prediction based on Forward-Backward View Transformation
- Global Vision Transformer Pruning with Hessian-Aware Saliency
- Heterogeneous Continual Learning
- Recurrence without Recurrence: Stable Video Landmark Detection with Deep Equilibrium Models
- The Best Defense is a Good Offense: Adversarial Augmentation Against Adversarial Attacks
- Zero-shot Pose Transfer for Unrigged Stylized 3D Characters
- Online Consistent Video Depth using Continuous Geometric Representations
- Pseudoinverse-Guided Diffusion Models for Inverse Problems
- Towards Annotation-efficient Segmentation via Image-to-image Translation
- LANA: Latency Aware Network Acceleration
- Neural Light Field Estimation for Outdoor Scenes with Differentiable Virtual Object Insertion
- Privacy Vulnerability of Split Computing to Data-Free Model Inversion Attacks
- RTMV: A Ray-Traced Multi-View Synthetic Dataset for Novel View Synthesis
- A-ViT: Adaptive Tokens for Efficient Vision Transformer
- CoordGAN: Self-Supervised Dense Correspondences Emerge from GANs
- FreeSOLO: Learning to Segment Objects without Annotations
- GLAMR: Global Occlusion-Aware Human Mesh Recovery with Dynamic Cameras
- GradViT: Gradient Inversion of Vision Transformers
- GroupViT: Semantic Segmentation Emerges From Text Supervision
- Physics Informed RNN-DCT Networks for Time-Dependent Partial Differential Equations
- Displacement-Invariant Cost Computation for Efficient Stereo Matching
- DRaCoN--Differentiable Rasterization Conditioned Neural Radiance Fields for Articulated Avatars
- Learning contrastive representation for semantic correspondence
- Do Gradient Inversion Attacks Make Federated Learning Unsafe?
- Neural Interferometry: Image Reconstruction from Astronomical Interferometers using Transformer-Conditioned Neural Fields
- A Contrastive Learning Approach for Training Variational Autoencoder Priors
- Coupled Segmentation and Edge Learning Using Dynamic Graph Propagation
- Coupled Segmentation and Edge Learning via Dynamic Graph Propagation
- KAMA: 3D Keypoint Aware Body Mesh Articulation
- Score-based Generative Modeling in Latent Space
- Hierarchical Contrastive Motion Learning for Video Action Recognition
- Learning Indoor Inverse Rendering with 3D Spatially-Varying Lighting
- NViT: Vision Transformer Compression and Parameter Redistribution
- Self-Supervised Object Detection via Generative Image Synthesis
- Domain Stylization: A Fast Covariance Matching Framework towards Domain Adaptation
- Adversarial Motion Modelling Helps Semi-Supervised Hand Pose Estimation
- Binary TTC: A Temporal Geofence for Autonomous Navigation
- DexYCB: A Benchmark for Capturing Hand Grasping of Objects
- Learning to Track Instances without Video Annotations
- See through Gradients: Image Batch Recovery via GradInversion
- Self-Supervised Learning on 3D Point Clouds by Learning Discrete Generative Models
- Weakly-Supervised Physically Unconstrained Gaze Estimation
- Learning continuous environment fields via implicit functions
- Parameter Efficient Multimodal Transformers for Video Representation Learning
- VAEBM: A Symbiosis between Variational Autoencoders and Energy-based Models
- Convolutional Tensor-Train LSTM for Spatio-temporal Learning
- NVAE: A Deep Hierarchical Variational Autoencoder
- Online adaptation for consistent mesh reconstruction in the wild
- Self-supervised single-view 3D reconstruction via semantic consistency
- Contrastive Learning for Weakly Supervised Phrase Grounding
- Joint Disentangling and Adaptation for Cross-Domain Person Re-Identification
- UFO2: A Unified Framework towards Omni-supervised Object Detection
- Angular Visual Hardness
- Bi3D: Stereo Depth Estimation via Binary Classifications
- Instance-aware, Context-focused, and Memory-efficient Weakly-Supervised Object Detection
- Meshlet Priors for 3D Mesh Reconstruction
- Novel View Synthesis of Dynamic Scenes With Globally Coherent Depths From a Monocular Camera
- Self-Supervised Viewpoint Learning from Image Collections
- Two-shot Spatially-varying BRDF and Shape Estimation
- UNAS: Differentiable Architecture Search Meets Reinforcement Learning
- Weakly-Supervised 3D Human Pose Learning via Multi-view Images in the Wild
- NRMVS: Non-Rigid Multi-View Stereo