Proposes an end-to-end architecture replacing standard CNN backbones with Swin Transformer to capture global image context. Introduces Multi-label Focal Loss to handle severe token imbalance in chemical datasets.
This paper presents a two-stage deep learning pipeline to extract chemical structures from documents and convert them to SMILES strings. By training on large-scale synthetic data, the method overcomes the brittleness of rule-based systems and demonstrates high accuracy even on low-resolution and noisy input images.
A 2019 format specification introducing two complementary standards for chemical mixtures. Mixfile provides comprehensive mixture descriptions and MInChI provides compact canonical identifiers. This addresses the long-standing lack of standardized machine-readable formats for multi-component chemical systems.
A 2018 infrastructure paper introducing RInChI (Reaction InChI), the first standardized format for uniquely identifying chemical reactions through algorithmic hashing and layering, enabling reaction database searching and duplicate detection analogous to how InChI works for individual molecules.
Discover how OCSR technology bridges the gap between molecular images and machine-readable data, evolving from rule-based systems to modern deep learning models for chemical knowledge extraction.
The MolParser project introduces two key datasets: MolParser-7M, the largest training dataset for Optical Chemical Structure Recognition (OCSR) with 7.7M pairs of images and E-SMILES strings, and WildMol, a new 20k-sample benchmark for evaluating models on challenging real-world data. The training data uniquely combines millions of diverse synthetic molecules with 400,000 manually annotated in-the-wild samples to significantly enhance model robustness.
A 2024 end-to-end OCSR system addressing both technical and data challenges, introducing MolParser-7M (7M+ image-text pairs) and MolDet (YOLO-based detector) for extracting and recognizing molecular structures from real-world documents with diverse quality and styles.
ZINC-22 is a multi-billion-scale public database containing over 37 billion make-on-demand molecules. It utilizes distributed infrastructure and specialized search algorithms to support modern ultra-large virtual screening campaigns.
MARCEL provides a comprehensive benchmark for molecular representation learning with 722K+ conformers across four diverse subsets (Drugs-75K, Kraken, EE, BDE), enabling evaluation of conformer ensemble methods for property prediction in drug discovery and catalysis.
GEOM contains 450k+ molecules with 37M+ conformations, featuring energy annotations from semi-empirical (GFN2-xTB) and DFT methods for property prediction and molecular generation research.
GDB-11 contains 26.4 million systematically generated small organic molecules with up to 11 atoms, establishing the methodology for exploring drug-like chemical space computationally.
GDB-13 contains nearly 1 billion systematically generated small organic molecules with up to 13 atoms, achieving billion-scale chemical space exploration while maintaining drug-like properties.