Skip to content

Latest commit

 

History

History
69 lines (52 loc) · 3.27 KB

README.md

File metadata and controls

69 lines (52 loc) · 3.27 KB

4D-LRM: Large Space-Time Reconstruction Model
From and To Any View at Any Time

Martin Ziqiao Ma, Xuweiyi Chen, Shoubin Yu, Sai Bi, Kai Zhang, Chen Ziwen, Sihan Xu, Jianing Yang, Zexiang Xu, Kalyan Sunkavalli, Mohit Bansal, Joyce Chai, Hao Tan

Adobe     UMich     UNC     UVA     OSU

Teaser

Paper Project Page Hugging Dataset

This repository contains a self-reimplemented (non-official!) version of 4D-LRM, including the model code, as well as training and evaluation pipelines. Part of the infrastructure is based on Long-LRM, credit to the amazing Ziwen Chen.

TODO List

Precommit Setup

This is for the authors (and contributors), you can skip this step.

We use Google docstring format for our docstrings and the pre-commit library to check our code. To install pre-commit, run the following command:

pip install pre-commit  # or conda install pre-commit
pre-commit install

The pre-commit hooks will run automatically when you try to commit changes to the repository.

Data Preparation

Model Configuration

Teaser

Model Training

torchrun --nproc_per_node $NUM_NODE --nnodes 1 \
         --rdzv_id $JOB_ID --rdzv_backend c10d --rdzv_endpoint localhost:$PORT \
         main.py --config path_to_your_config.yaml \
         --default-config path_to_your_default_config.yaml

Model Evaluation

To run the evaluation loop, add the --evaluation flag to the command line:

torchrun --nproc_per_node $NUM_NODE --nnodes 1 \
         --rdzv_id $JOB_ID --rdzv_backend c10d --rdzv_endpoint localhost:$PORT \
         main.py --config path_to_your_config.yaml \
         --default-config path_to_your_default_config.yaml \
         --evaluation