HAPC-OCSR

OCSR implemented by Hallym APCLab

Overview

HAPC-OCSR is an Optical Chemical Structure Recognition (OCSR) system developed by the AI-Powered Cheminformatics Laboratory (APCLab), Hallym University.
The tool provides a Tkinter-based GUI to:

• Upload a chemical structure image.
• Predict the corresponding SMILES string using a PyTorch model.
• Visualize both the uploaded image and the RDKit-rendered molecular structure.
• Copy the predicted SMILES string with a single click.

---

Installation

Note for Windows users
Since this repository provides a Windows-specific launcher (run_app.bat), please place the HAPC-OCSR-master directory under your Documents folder:

C:\Users\%USERNAME%\Documents\HAPC-OCSR-master

1. Clone the repository

   git clone https://github.com/mathcom/HAPC-OCSR.git
   cd HAPC-OCSR

2. Create and activate conda environment

   conda env create -f environment.yml
   conda activate hapc-ocsr

3. Download model checkpoint

   • Create a folder named ckpt in the project root:

     mkdir ckpt

   • Download the pretrained models from Google Drive:
     molscribe+ocsaug.pth
     molnextr+ocsaug.pth
   • Place the files into the ckpt folder:

     HAPC-OCSR-master/
     ├── main.py
     ├── run_app.bat
     ├── environment.yml
     ├── ckpt/
     │   └── molscribe+ocsaug.pth
     │   └── molnextr+ocsaug.pth
     

---

Usage

Option 1: Run with Conda

python main.py

Option 2: Run with Batch file (Windows only)

Simply double-click:

run_app.bat

This script will automatically launch the program using:

%USERPROFILE%\miniconda3\envs\hapc-ocsr\python.exe %USERPROFILE%\Documents\HAPC-OCSR-master\main.py

---

Acknowledgements

• MolScribe and MolNexTR for OCSR backbone.
• RDKit, Pillow, and Tkinter for visualization and GUI.

---

Citation

If you use HAPC-OCSR in your research, please cite the following papers:

• Kim, Jin Hyuk, and Jonghwan Choi. "OCSAug: diffusion-based optical chemical structure data augmentation for improved hand-drawn chemical structure image recognition." The Journal of Supercomputing 81.8 (2025): 926. https://doi.org/10.1007/s11227-025-07406-4
• Chen, Yufan, et al. "MolNexTR: a generalized deep learning model for molecular image recognition." Journal of Cheminformatics 16.1 (2024): 141. https://doi.org/10.1186/s13321-024-00926-w
• Qian, Yujie, et al. "MolScribe: robust molecular structure recognition with image-to-graph generation." Journal of Chemical Information and Modeling 63.7 (2023): 1925-1934. https://doi.org/10.1021/acs.jcim.2c01480

---

Contact

jonghwanc@hallym.ac.kr