Welcome to the Pytorch Implementation of "ActMVS: Active Scene Reconstruction with Monocular Multi-View Stereo" for ICRA 2026.

0.Setup

Create Conda Env

conda create -y -n ActMVS python=3.9 cmake=3.14.0
conda activate ActMVS

pip install torch==2.1.2 torchvision==0.16.2 torchaudio==2.1.2 --index-url https://download.pytorch.org/whl/cu118
pip install git+https://github.com/liren-jin/diff-gaussian-rasterization_2d
pip install git+https://github.com/nianticlabs/mvsanywhere

cd simulator
git clone git@github.com:liren-jin/habitat-sim.git
cd habitat-sim
pip install -r requirements.txt
python setup.py install --headless --bullet

Download Replica Data

bash scripts/replica_download.sh

1.Run

make sure that "mesh_path" and "scene_id" in "config.yaml" are pointing to your downloaded replica data.

bash scripts/run.sh

Feel free to change hyperparameters in "config.yaml". As default, "budget: 600" sets the total exploration duration to 600 seconds (increase for more completion).

2.Check Results

Intermediate visualizations during active reconstruction:

From left to right: Input Frame, Predicted Depth, Rendered Mask, Rendered Frame, Rendered Depth, Rendered Normal, Rendered Opacity.

Intermediate Point Clouds:

Final Mesh:

Point clouds and Mesh are stored as ".ply" files and can be viewed in MeshLab.

Thanks for your attention! More instructions will be added soon!