MolClassifier is a model to classify chemical images into categories: "Clean" (corresponding to Molecular Structure), "Markush" (corresponding to Markush structure) and "Trash" (corresponding to Background). MolClassifier is presented in PatCID: an open-access dataset of chemical structures in patent documents.
Ensure your distribution has Python version 3.11:
python3 --version
pip3 -V
Install poetry
pip install poetry
Install the dependencies
poetry install
wget https://huggingface.co/ds4sd/MolClassifier/resolve/main/models/molclassifier_model.chpt -P ./data/models/
-
Place the images to annotate in
./data/dataset/ -
Run the model
poetry run python ./mol_classifier/classifier.py infer --dir ./data/dataset/ --checkpoint ./data/models/molclassifier_model.chpt --output ./data/output.txt -
Read predictions in
./data/output.txt
The training and validation datasets are available on Zenodo.