Cosmos is a world model development platform that consists of world foundation models, tokenizers and video processing pipeline to accelerate the development of Physical AI at Robotics & AV labs. C…

Official repo and evaluation implementation of VSI-Bench

Official implementation of the paper “MagicDriveDiT: High-Resolution Long Video Generation for Autonomous Driving with Adaptive Control”

Official implementation of our RAL'24 paper: Multi-Camera Unified Pre-training for Autonomous Driving

A generative world for general-purpose robotics & embodied AI learning.

Code for the project "MegaSaM: Accurate, Fast and Robust Structure and Motion from Casual Dynamic Videos"

Official implementation of the paper "HUGSIM: A Real-Time, Photo-Realistic and Closed-Loop Simulator for Autonomous Driving"

[SIGGRAPH Asia 2024 (Journal Track)] StableNormal: Reducing Diffusion Variance for Stable and Sharp Normal

[ECCV2024] Gaussian Splatting on the Move: Blur and Rolling Shutter Compensation for Natural Camera Motion

SplatAD: Real-Time Lidar and Camera Rendering with 3D Gaussian Splatting for Autonomous Driving

Official Implementation of Driv3R

HunyuanVideo: A Systematic Framework For Large Video Generation Model

Code release for https://kovenyu.com/WonderWorld/

Official repository of "SAMURAI: Adapting Segment Anything Model for Zero-Shot Visual Tracking with Motion-Aware Memory"

[CVPR 2024 Highlight] XCube: Large-Scale 3D Generative Modeling using Sparse Voxel Hierarchies

Qwen2.5 is the large language model series developed by Qwen team, Alibaba Cloud.

Official Implementation of paper "MonST3R: A Simple Approach for Estimating Geometry in the Presence of Motion"

GVKF: Gaussian Voxel Kernel Functions for Highly Efficient Surface Reconstruction in Open Scenes

[NeurIPS 2024] SCube: Instant Large-Scale Scene Reconstruction using VoxSplats

Implementation of ECCV'24: GaussReg: Fast 3D Registration with Gaussian Splatting

[NeurIPS 2024] ODGS: 3D Scene Reconstruction from Omnidirectional Images with 3D Gaussian Splatting

[arXiv'24] [Image-to-Scene on a 4090(24G)] VistaDream: Sampling multiview consistent images for single-view scene reconstruction

FreeVS: Generative View Synthesis on Free Driving Trajectory

DepthSplat: Connecting Gaussian Splatting and Depth

Depth Any Video with Scalable Synthetic Data