Tailored Refinement of Vision-Language Models for Plant Instance Segmentation

Paper   •   Contact Us

Setup

Build docker image:

make build

You may need to install the NVIDIA container toolkit to bring the GPUs to work with Docker.

Data Samples and Weights

We provide a test folder containing a subset of images from PhenoBench with their semantic and plant instance labels. This allows to check the pipeline. To perform a full validation, one can download the PhenoBench dataset from https://www.phenobench.org/.

Option 1: Manual download

Download the data samples here: samples.zip,

Unzip the samples and copy the folder samples and the weights into the main folder.

Option 2: Automated download

Execute

make download

Test on samples

We provide some samples to try the label refinement. You can run

make test

which will also save the result as one image with 4 plots: input, semantic label, instance label, and instance prediction.

How make the code your own

Data

If you want to use other data

1. In the makefile change the DATA_PATH to point at your data
2. Write or import the dataloader: a. Implement the dataloader in the datasets folder b. change the init.py file in the datasets folder to import the dataloader c. Change your config file data name and root_dir to access your new data

Starting from another coarse instance segmentation

1. Load the coarse instance segmentation in your dataloader
2. Change the config file to use the coarse instance segmentation (not the color based one)

Style Guidelines

In general, we follow the Python PEP 8 style guidelines. Please install black to format your python code properly. To run the black code formatter, use the following command:

black -l 120 path/to/python/module/or/package/

To optimize and clean up your imports, feel free to have a look at this solution for PyCharm.