google/siglip2-so400m-patch16-256

Note

This is a remotely-sourced model from the siglip2 plugin, maintained by the community. It is not part of FiftyOne core and may have special installation requirements. Please review the plugin documentation and license before use.

SigLIP 2 extends the pretraining objective of SigLIP with prior, independently developed techniques into a unified recipe, for improved semantic understanding, localization, and dense features..

Details

• Model name: google/siglip2-so400m-patch16-256

• Model source: https://huggingface.co/google/siglip2-so400m-patch16-256

• Model author: Google

• Model license: Apache-2.0

• Exposes embeddings? yes

• Tags: classification, logits, embeddings, torch, clip, zero-shot

Requirements

• Packages: huggingface-hub, transformers, torch, torchvision, accelerate

• CPU support

  • yes

• GPU support

  • yes

Example usage

import fiftyone as fo
import fiftyone.zoo as foz

foz.register_zoo_model_source("https://github.com/harpreetsahota204/siglip2")

dataset = foz.load_zoo_dataset(
    "coco-2017",
    split="validation",
    dataset_name=fo.get_default_dataset_name(),
    max_samples=50,
    shuffle=True,
)

model = foz.load_zoo_model("google/siglip2-so400m-patch16-256")

dataset.apply_model(model, label_field="predictions")

session = fo.launch_app(dataset)