Awesome
<div align="center">๐งช DECIMER Image Transformer ๐ผ๏ธ
Deep Learning for Chemical Image Recognition using Efficient-Net V2 + Transformer
<p align="center"> <img src="https://github.com/Kohulan/DECIMER-Image_Transformer/blob/master/DECIMER_V2.png?raw=true" alt="DECIMER Logo" width="600"> </p> </div>๐ Table of Contents
- Abstract
- Method and Model Changes
- Installation
- Usage
- Hand-drawn Model
- Citation
- Acknowledgements
- Author
- Project Website
- Research Group
๐ฌ Abstract
<div style="background-color: #f0f0f0; padding: 15px; border-radius: 10px;">The DECIMER 2.2 project tackles the OCSR (Optical Chemical Structure Recognition) challenge using cutting-edge computational intelligence methods. Our goal? To provide an automated, open-source software solution for chemical image recognition.
We've supercharged DECIMER with Google's TPU (Tensor Processing Unit) to handle datasets of over 1 million images with lightning speed!
</div>๐ง Method and Model Changes
<table> <tr> <td width="50%"> <h3>๐ผ๏ธ Image Feature Extraction</h3> <p>Now utilizing EfficientNet-V2 for superior image analysis</p> </td> <td width="50%"> <h3>๐ฎ SMILES Prediction</h3> <p>Employing a state-of-the-art transformer model</p> </td> </tr> </table>๐ Training Enhancements
- TFRecord Files: Lightning-fast data reading
- Google Cloud Buckets: Efficient cloud storage solution
- TensorFlow Data Pipeline: Optimized data loading
- TPU Strategy: Harnessing the power of Google's TPUs
๐ป Installation
# Create a conda wonderland
conda create --name DECIMER python=3.10.0 -y
conda activate DECIMER
# Equip yourself with DECIMER
pip install decimer
๐ฎ Usage
from DECIMER import predict_SMILES
# Unleash the power of DECIMER
image_path = "path/to/your/chemical/masterpiece.jpg"
SMILES = predict_SMILES(image_path)
print(f"๐ Decoded SMILES: {SMILES}")
โ๏ธ DECIMER - Hand-drawn Model
<div style="background-color: #e6f7ff; padding: 15px; border-radius: 10px;">๐ New Feature Alert! ๐
Our latest model brings the magic of AI to hand-drawn chemical structures!
</div>๐ Citation
<div style="background-color: #f9f9f9; padding: 15px; border-radius: 10px;">If DECIMER helps your research, please cite:
- Rajan K, et al. "DECIMER.ai - An open platform for automated optical chemical structure identification, segmentation and recognition in scientific publications." Nat. Commun. 14, 5045 (2023).
- Rajan, K., et al. "DECIMER 1.0: deep learning for chemical image recognition using transformers." J Cheminform 13, 61 (2021).
- Rajan, K., et al. "Advancements in hand-drawn chemical structure recognition through an enhanced DECIMER architecture," J Cheminform 16, 78 (2024).
๐ Acknowledgements
- A big thank you to Charles Tapley Hoyt for his invaluable contributions!
- Powered by Google's TPU Research Cloud (TRC)
๐จโ๐ฌ Author: Kohulan
<p align="center"> <img src="https://github.com/Kohulan/DECIMER-Image-to-SMILES/raw/master/assets/DECIMER.gif" width="300"> </p>๐ Project Website
Experience DECIMER in action at decimer.ai, brilliantly implemented by Otto Brinkhaus!
๐ซ Research Group
<p align="center"> <a href="https://cheminf.uni-jena.de"> <img src="https://github.com/Kohulan/DECIMER-Image-to-SMILES/blob/master/assets/CheminfGit.png" width="300"> </a> </p><div align="center">