Kohulan / DECIMER-Image_Transformer

DECIMER: Deep Learning for Chemical Image Recognition using Efficient-Net V2 + Transformer
MIT License
197 stars 51 forks source link

How to get the positional informations of the molecule #77

Closed Liaokang123456 closed 12 months ago

Liaokang123456 commented 12 months ago

Issue Type

Questions

Source

GitHub (source)

DECIMER Image Transformer Version

2.0

OS Platform and Distribution

No response

Python version

No response

Current Behaviour?

from DECIMER import predict_SMILES

Chemical depiction to SMILES translation

image_path = "path/to/imagefile" SMILES = predict_SMILES(image_path) print(SMILES)

The SMILES do not have the atom coordinates. How to get the positional information? Thanks!

Which images caused the issue? (This is mandatory for images related issues)

No response

Standalone code to reproduce the issue

from DECIMER import predict_SMILES

# Chemical depiction to SMILES translation
image_path = "path/to/imagefile"
SMILES = predict_SMILES(image_path)
print(SMILES)

Relevant log output

No response

Code of Conduct

OBrink commented 12 months ago

Hey @Liaokang123456,

Thank you for your interest in DECIMER! We are currently training a model that is capable of generating cxSMILES with atom coordinates. It will appear here in the repository in the upcoming months after training and validation are done. The current version only generates a SMILES string without the positional information of the atoms in the image.

If you need an OCSR tool that is capable of generating the atom coordinates now, maybe have a look at MolScribe.

Kind regards, Otto