MolAnnotator

This repository implements applications to ease manual annotation of chemical images, used in PatCID: an open-access dataset of chemical structures in patent documents.

Installation

conda create -n molannotator python=3.11
conda activate molannotator
pip install -e .

Classification

The classification application allows to assign to an image one of the following categories: standard_molecule (Molecular Structure), markush_structure (Markush Structure), miscellaneous (Background), polymer (Markush Structure), and segmentation_annotation_error.

To run the application:

1. Move the images to be annotated to ./data/classfication/default/images/*.png,
2. (Optional) Move pre-annotations to ./data/classfication/default/classes/*.json,
3. Run bash ./molannotator/apps/run_classification_app.sh.

Annotations are saved in ./data/classfication/default/classes/*.json.

Recognition

The recognition application allows to annotate the molecular graph of a chemical-structure image.

To run the application:

1. Move the images to be annotated to ./data/recognition/default/images/*.png,
2. (Optional) Create pre-annotations by running python3 ./molannotator/molscribe_pre_annotate.py (see MolScribe for installation steps),
3. (Optional) Move pre-annotations to ./data/recognition/default/predictions/*.mol,
4. Run bash ./molannotator/apps/run_recognition_app.sh.

Annotations are saved in ./data/recognition/default/molfiles/*.mol.