BENBV (Boundary Exploration Next Best View)

This is the official repository for the Boundary Exploration of Next Best View Policy in 3D Robotic Scanning.

Environment

The conda is recommended using to create a virtual environment as:

conda create -n nbv python=3.8 ipython
conda activate nbv

First, the Pytorch is used in nbv_explore_net such that you can install Pytorch as:

(depends on the CUDA you installed)
pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 --index-url https://download.pytorch.org/whl/cu121

The cpu version for Pytorch is also available if you just want to run the code in dataset_generation.

Second, the other packages are required to install as:

pip install -r requirements-no-pytorch.txt

The version of Pytorch 2.3.1 and Pytorch 1.8 has been tested.

Usage

1. Dataset Generation (also BENBV included)

• Jump to ./dataset_generation and run python main.py
• Stay and run python show_views.py where you can select the next-best-view visualization.

2. BENBV-Net

• Jump to ./nbv_explore_net and run python train.py