Euclid Quick Data Release (Q1). AstroVink: A vision transformer approach to find strong gravitational lens systems
Paper • 2604.21977 • Published
AstroVink (Q1 version)
This model card describes AstroVink-Q1, a Vision Transformer-based model for automated detection of strong gravitational lens systems in Euclid Space Telescope imaging data.
Model Description
AstroVink-Q1 is a Vision Transformer (ViT-S/14) model with a DINOv2 backbone, fine-tuned for binary image classification of strong gravitational lens candidates in Euclid Q1 survey data. The model predicts lens vs non-lens probabilities from cutout images.
- Developed by: Saamie Vincken
- Model type: Vision Transformer (ViT)
- Finetuned from model: DINOv2 ViT-S/14
- License: MIT
- Repository: https://github.com/SaamieVincken/AstroVink
- Paper: https://arxiv.org/abs/2604.21977
- Datasets: https://zenodo.org/records/15025832 and https://zenodo.org/records/17425610
When using, please cite the AstroVink paper and any related papers to the Strong Lensing Discovery Engine of the Euclid Q1 data release.
Astrovink Citations
Bibtex:
@article{vincken2026astrovink, title={Euclid Quick Data Release (Q1). AstroVink: A vision transformer approach to find strong gravitational lens systems}, author={Vincken, Saamie H.}, journal={Astronomy & Astrophysics}, year={2026}, note={in preparation}, url={https://arxiv.org/abs/2604.21977} }
APA:
Vincken, S. H. (2026). Euclid Quick Data Release (Q1). AstroVink: A vision transformer approach to find strong gravitational lens systems. Astronomy & Astrophysics (in preparation). https://arxiv.org/abs/2604.21977
Model Card Contact
Saamie Vincken (saamievincken@protonmail.com)