Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Related Concept Videos

Molecular Models

Molecular Models

Physical models representing molecular architectures of chemical compounds play essential roles in understanding chemistry. The use of molecular models makes it easier to visualize the structures and shapes of atoms and molecules.

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

Encapsulated droplet array for intensified micromole-scale high-throughput reaction screening.

Nature communications·2026

Same author

Machine learning-driven multi-objective optimization of electrospun nanofibrous membranes design for membrane distillation.

Water research·2026

Same author

Inhibitory Effect of <i>Allium cepa L</i>.-Derived Extracellular Vesicles Loaded with Celecoxib on Osteoclast Differentiation in Periodontitis.

International journal of nanomedicine·2026

Same author

High-capacity optical communication relayed by multi-core amplifier on deployed submarine seven-core fiber cable.

Communications engineering·2026

Same author

MG-3D: Multi-grained knowledge-enhanced vision-language pre-training for 3D medical image analysis.

Medical image analysis·2026

Same author

REM Sleep Behavior Disorder: Pathophysiology, Biomarkers, and Therapeutic Strategies.

Journal of integrative neuroscience·2026

Search research articles

Related Experiment Video

Updated: May 16, 2026

Deep Learning-Based Segmentation of Cryo-Electron Tomograms

Deep Learning-Based Segmentation of Cryo-Electron Tomograms

Published on: November 11, 2022

MolNexTR: a generalized deep learning model for molecular image recognition.

Yufan Chen¹, Ching Ting Leung¹, Yong Huang²

¹Department of Chemical and Biological Engineering, Hong Kong University of Science and Technology, Hong Kong, SAR, China.

Journal of Cheminformatics

|December 19, 2024

Summary

This summary is machine-generated.

MolNexTR, a new deep learning model, accurately converts molecular images to machine-readable data. This advanced image-to-graph approach enhances chemical structure recognition by fusing ConvNext and Vision-Transformer strengths.

Keywords:

Chemical structure recognition ConvNext Deep learning Transformer

More Related Videos

DNA Virus Detection System Based on RPA-CRISPR/Cas12a-SPM and Deep Learning

DNA Virus Detection System Based on RPA-CRISPR/Cas12a-SPM and Deep Learning

Published on: May 10, 2024

Analyzing Mitochondrial Morphology Through Simulation Supervised Learning

Analyzing Mitochondrial Morphology Through Simulation Supervised Learning

Published on: March 3, 2023

Related Experiment Videos

Last Updated: May 16, 2026

Deep Learning-Based Segmentation of Cryo-Electron Tomograms

Deep Learning-Based Segmentation of Cryo-Electron Tomograms

Published on: November 11, 2022

DNA Virus Detection System Based on RPA-CRISPR/Cas12a-SPM and Deep Learning

DNA Virus Detection System Based on RPA-CRISPR/Cas12a-SPM and Deep Learning

Published on: May 10, 2024

Analyzing Mitochondrial Morphology Through Simulation Supervised Learning

Analyzing Mitochondrial Morphology Through Simulation Supervised Learning

Published on: March 3, 2023

Area of Science:

Computational chemistry
Cheminformatics
Deep learning for chemical structure recognition

Background:

Chemical structure recognition from images is challenging due to varied drawing styles.
Converting molecular images to machine-readable formats like SMILES strings is crucial for data analysis.

Purpose of the Study:

To introduce MolNexTR, a novel image-to-graph deep learning model for accurate molecular image recognition.
To improve the extraction of local and global features from molecular images.
To enhance the model's robustness to diverse molecular image styles.

Main Methods:

Developed MolNexTR, an image-to-graph deep learning model fusing ConvNext and Vision-Transformer.
Integrated symbolic chemistry principles for chirality and abbreviated structure recognition.
Employed advanced algorithms: improved data augmentation, image contamination, and post-processing modules.

Main Results:

MolNexTR achieved superior performance in molecular structure recognition tasks.
Accuracy rates ranged from 81% to 97% on test sets.
Demonstrated effective prediction of atoms, bonds, and their layout rules.

Conclusions:

MolNexTR represents a significant advancement in molecular structure recognition.
The model's dual-stream encoder and integrated chemical rules enhance feature extraction and prediction accuracy.
Novel augmentation algorithms contribute to improved robustness and performance on diverse molecular images.