MolParser-7M

# MolParser-7M [**Demo**](https://ocsr.dp.tech/) | [**Paper**](https://arxiv.org/abs/2411.11098) | [**Code**](https://github.com/dptech-corp/MolParser) This repo provides the training data and evaluation data for MolParser, proposed in paper *“MolParser: End-to-end Visual Recognition of Molecule Structures in the Wild“* (ICCV2025 accept) MolParser-7M contains nearly 8 million paired image-SMILES data. It should be noted that the caption of image is our extended-SMILES format, which suggested in our paper. * **MolParser-7M (Pretrain)**: More than 7.7M synthetic training data in `pretrain_synthetic_7M` subset; * **MolParser-SFT**: Human-labeled real molecule figures for fine-tuning stage in `sft_real` subset. (We are organizing an OCSR competition based on MolParser-7M, so we have reserved part of the MolParser-SFT data for the competition. Stay tuned!) * **MolParser-Val**: A small validation set carefully selected in-the-wild in `valid` subset. It can be used to quickly valid the model ability during the training process; * **WildMol Benchmark**: 20k molecule structure images cropped from real patents or paper, `test_simple_10k`（WildMol-10k）subset and `test_markush_10k`（WildMol-10k-M）subset; ## 📜 License This dataset is provided for **non-commercial use only**. For commercial use, please contact: [[e-mail]](mailto:[e-mail]) or add a discussion in HuggingFace. ## 📖 Citation If you use this datasets in your work, please cite: ``` @inproceedings{fang2025molparser, title={Molparser: End-to-end visual recognition of molecule structures in the wild}, author={Fang, Xi and Wang, Jiankun and Cai, Xiaochen and Chen, Shangqian and Yang, Shuwen and Tao, Haoyi and Wang, Nan and Yao, Lin and Zhang, Linfeng and Ke, Guolin}, booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision}, pages={24528--24538}, year={2025} } ```