NVIDIA Research Projects

Eagle: Frontier Vision-Language Models with Data-Centric Strategies

[CVPR 2026] Fast-FoundationStereo: Real-Time Zero-Shot Stereo Matching

Gradient Boosting Reinforcement Learning (GBRL)

[CVPR 2025 Best Paper Nomination] FoundationStereo: Zero-Shot Stereo Matching

[ICCV 2025] Implementation for Describe Anything: Detailed Localized Image and Video Captioning

Long-RL: Scaling RL to Long Sequences (NeurIPS 2025)

Official implementation of "Neuralangelo: High-Fidelity Neural Surface Reconstruction" (CVPR 2023)

ToolOrchestra is an end-to-end RL training framework for orchestrating tools and agentic workflows.

[ICLR 2025] Official PyTorch Implementation of Gated Delta Networks: Improving Mamba2 with Delta Rule

[ICLR 2026] LongLive: Real-time Interactive Long Video Generation

We propose a novel modular framework that learns to dynamically mix low-rank adapters (LoRAs) to improve visual analogy learning, enabling flexible and generalizable image edits based on example transformations.

Welcome to GR00T Whole-Body Control (WBC)! This is a unified platform for developing and deploying advanced humanoid controllers. This includes: Decoupled WBC models used in NVIDIA Isaac-Gr00t, Gr00t N1.5 and N1.6 and GEAR-SONIC

[CVPR 2024 Highlight] FoundationPose: Unified 6D Pose Estimation and Tracking of Novel Objects

Instant neural graphics primitives: lightning fast NeRF and more

[CVPR 2026] CARI4D: Category Agnostic 4D Reconstruction of Human-Object Interaction

[ICLR 2026] rCM: SOTA JVP-Based Diffusion Distillation & Few-Step Video Generation & Scaling Up sCM/MeanFlow & Real-Time Autoregressive Video Diffusion

SANA: Efficient High-Resolution Image Synthesis with Linear Diffusion Transformer

OmniVinci is an omni-modal LLM for joint understanding of vision, audio, and language.

Make your wildest 3D ConvNet dream architectures come true

Data Generation Pipeline for Mobility

[ICML2024 (Oral)] Official PyTorch implementation of DoRA: Weight-Decomposed Low-Rank Adaptation

Cosmos Policy

StyleGAN - Official TensorFlow Implementation

NVIDIA FastGen: Fast Generation from Diffusion Models

Official implementation of GDPO: Group reward-Decoupled Normalization Policy Optimization for Multi-reward RL Optimization

CUDA Accelerated Robot Library

[CoRL 2025] Neural Robot Dynamics