You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Our Occupancy World Model can generate long-duration occupancy forecasts and can be effectively controlled by trajectory conditions.
π Overview
Our method consists of two components: (a) Occ-VAE Pipeline encodes occupancy frames into a continuous latent space, enabling efficient data compression. (b)DOME Pipeline learns to predict 4D occupancy based on historical occupancy observations.
ποΈ News
[2025.1.1] We release the code and checkpoints.
[2024.11.18] Project page is online!
ποΈ TODO
Code release.
Checkpoint release.
π Setup
clone the repo
git clone https://github.com/gusongen/DOME.git
cd DOME
environment setup
conda env create --file environment.yml
data preparation
Create soft link from data/nuscenes to your_nuscenes_path
Prepare the gts semantic occupancy introduced in Occ3d
Download our generated train/val pickle files and put them in data/
@article{gu2024dome,
title={Dome: Taming diffusion model into high-fidelity controllable occupancy world model},
author={Gu, Songen and Yin, Wei and Jin, Bu and Guo, Xiaoyang and Wang, Junming and Li, Haodong and Zhang, Qian and Long, Xiaoxiao},
journal={arXiv preprint arXiv:2410.10429},
year={2024}
}
About
official code of *DOME: Taming Diffusion Model into High-Fidelity Controllable Occupancy World Model*