Co-MOF

Environment

They can be installed using conda or a requirements.txt file.

Using the conda environment file

1. Install conda
2. Create a new environment using conda env create -f environment/conda/environment.linux.yml

Using the requirements.txt file

1. Use pip install -r environment/requirements.txt to install dependencies. If creating your own environment, python >= 3.10 is recommended.

Dataset

The dataset is located in co-mof_dataset/, and contains images as well as supporting files

In images/, there are a total of 796 images organized by batches. Each batch follows a specific procedure defined in batches.json. The procedures in batches.json are defined in the procedures.json file. mof_dataset.csv contains the relative image paths relative to the Co-MOF/ directory, as well as the procedure, batch, and synthesis condition for each image. Synthesis conditions are defined in synthesis_conditions.json.

When cloning this repository from GitHub, most images will be missing from images/. The remaining images can be downloaded TODO: HERE. After downloading all the images, run the code in the unit_testing.ipynb notebook. This notebook will load mof_dataset.csv into a pandas DataFrame, and attempt to open and close each image in the dataset to verify the image files' integrity as well as the CSV file.

Example Usage

The example_usage.ipynb demonstrates how to use Bok Choy with an optical microscopy image of MOFs. Using Bok Choy, results can be saved to a CSV file or plotted.

Expected Results

Image/sample quality has a significant impact on areas and aspect ratios determined by the Bok Choy algorithm.

Here are some examples of images with good, reasonable, and bad expected results:

Good Results Expected

Reasonable Results Expected

Bad Results Expected

Citation

If you use this dataset or Bok Choy in your research, please cite the following:

BibTeX

@article{Chong2025,
  title = {Controlling metal–organic framework crystallization via computer vision and robotic handling},
  ISSN = {2050-7496},
  url = {http://dx.doi.org/10.1039/D5TA03199K},
  DOI = {10.1039/d5ta03199k},
  journal = {Journal of Materials Chemistry A},
  publisher = {Royal Society of Chemistry (RSC)},
  author = {Chong,  Kai Xiang and Alsabia,  Qusai Abdulkhaliq and Ye,  Zuyang and McDaniel,  Andrew and Baumgardner,  Douglas and Xiao,  Dianne and Sun,  Shijing},
  year = {2025}
}