TreeSim_Lpy

Description

TreeSim_Lpy is a tree modelling tool which is built upon L-py with the added features of pruning and tying trees down to mimic different tree architectures such as Upright fruiting offshoot, V-Trellis, etc.

Applications

Learning to Prune Branches in Modern Tree-Fruit Orchard - https://ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=11128361

A Dataset for Semantic and Instance Segmentation of Modern Fruit Orchards - https://github.com/tqosu/MFO

Documentation

The documentation is provided at `Read the Docs <https://osurobotics.github.io/lpy_treesim/>

You can find the latest L-Py documentation at <https://lpy.readthedocs.io/en/latest>

Docker GUI (noVNC) and activating the lpy environment

You can run TreeSim inside a Docker container that provides a virtual X display and noVNC (a browser VNC client). This container also provides a conda environment named lpy containing the required LPy packages.

Build the Docker image (recommended tag: lpy-treesim:ubuntu22):

docker build -t lpy-treesim:ubuntu22 .

Run the container (detached), mounting your local cloned workspace:

# Mount the entire local repo to /app for editable install
docker run -d --name lpy-gui -p 6080:6080 -p 5900:5900 -v ${PWD}:/app lpy-treesim:ubuntu22

This mounts your local cloned workspace (with all files like .git, README.rst, etc.) to /app. The entrypoint installs the package in editable mode from the mounted code, so local changes are reflected immediately without rebuilding.

Connect to the GUI with your browser:

• Visit: http://localhost:6080/vnc.html?host=localhost&port=6080
• A desktop session with fluxbox and Xvfb will start, along with an xterm terminal.

To verify the environment is active inside the container:

# Should print 'lpy' if activated
echo $CONDA_DEFAULT_ENV
python -c "import openalea.lpy; print('openalea.lpy import ok')"

NOTE: You can copy and paste content in noVNC using the clipboard menu in the left bar and middle click (mouse button) to paste in the terminal.

See the repository Dockerfile for more details about the installed packages and build steps.

Making your first tree

To create your first tree using TreeSim_Lpy, in the NoVNC window with the terminal with L-Py activated, run the following commands:

python treesim_lpy/tree_generation/make_n_trees.py --num_trees 1 --output_dir ./dataset

This should create a dataset folder in the current directory with one generated tree mesh in it. It should be available locally on your host machine in the cloned repository folder.

Labeling Options

You can specify labeling options via command-line flags. Only one labeling mode can be enabled at a time:

# Enable semantic labeling
python lpy_treesim/tree_generation/make_n_trees.py --num_trees 1 --semantic-label

# Enable instance labeling
python lpy_treesim/tree_generation/make_n_trees.py --num_trees 1 --instance-label

# Enable per-cylinder labeling
python lpy_treesim/tree_generation/make_n_trees.py --num_trees 1 --per-cylinder-label

If no labeling flag is provided, no labeling is applied to the generated trees.

Features

• Pruning: Remove unwanted branches to simulate pruning.
• Branch Tying: Simulate branches being tied down to mimic different orchard architectures.
• Labelling: Get instance and semantic segmentation labels of the mesh

Contact

Please open an Issue if you need support or you run into any error (Installation, Runtime, etc.). We'll try to resolve it as soon as possible.