Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

SigLIP 2: Multilingual Vision-Language Encoders with Improved Semantic Understanding, Localization, and Dense Features

About

We introduce SigLIP 2, a family of new multilingual vision-language encoders that build on the success of the original SigLIP. In this second iteration, we extend the original image-text training objective with several prior, independently developed techniques into a unified recipe -- this includes captioning-based pretraining, self-supervised losses (self-distillation, masked prediction) and online data curation. With these changes, SigLIP 2 models outperform their SigLIP counterparts at all model scales in core capabilities, including zero-shot classification, image-text retrieval, and transfer performance when extracting visual representations for Vision-Language Models (VLMs). Furthermore, the new training recipe leads to significant improvements on localization and dense prediction tasks. We also train variants which support multiple resolutions and preserve the input's native aspect ratio. Finally, we train on a more diverse data-mixture that includes de-biasing techniques, leading to much better multilingual understanding and improved fairness. To allow users to trade off inference cost with performance, we release model checkpoints at four sizes: ViT-B (86M), L (303M), So400m (400M), and g (1B).

Michael Tschannen, Alexey Gritsenko, Xiao Wang, Muhammad Ferjad Naeem, Ibrahim Alabdulmohsin, Nikhil Parthasarathy, Talfan Evans, Lucas Beyer, Ye Xia, Basil Mustafa, Olivier H\'enaff, Jeremiah Harmsen, Andreas Steiner, Xiaohua Zhai• 2025

Related benchmarks

TaskDatasetResultRank
Semantic segmentationADE20K (val)
mIoU42.7
2731
Visual Question AnsweringTextVQA
Accuracy74.3
1117
Semantic segmentationADE20K
mIoU51.6
936
Image ClassificationImageNet-1K
Top-1 Acc88
836
Image ClassificationImageNet A
Top-1 Acc90.5
553
Image ClassificationImageNet-1K--
524
Image ClassificationFood-101
Accuracy97
494
Image ClassificationImageNet V2--
487
Image ClassificationStanford Cars
Accuracy95.9
477
Text-to-Image RetrievalFlickr30K
R@182.5
460
Showing 10 of 192 rows
...

Other info

Follow for update