This repository contains the code supporting the FastViT base model for use with Autodistill.

FastViT, developed by Apple, is a classification model that supports zero-shot classification.

Read the full Autodistill documentation.

Read the FastViT Autodistill documentation.

To use FastViT with autodistill, you need to install the following dependency:

pip3 install autodistill-fastvit

FastViT works using the ImageNet-1k class list. This class list is available in the FASTVIT_IMAGENET_1K_CLASSES variable.

You can provide classes from the list to retrieve predictions for a specific class in the list. You can also provide a custom ontology to map classes from the list to your own classes.

from autodistill_fastvit import FastViT, FASTVIT_IMAGENET_1K_CLASSES
from autodistill.detection import CaptionOntology

# zero shot with no prompts
base_model = FastViT(None)

# zero shot with prompts from FASTVIT_IMAGENET_1K_CLASSES
base_model = FastViT(
    ontology=CaptionOntology(
        {
            "coffeemaker": "coffeemaker",
            "ice cream": "ice cream"
        }
    )
)

predictions = base_model.predict("./example.png")

labels = [FASTVIT_IMAGENET_1K_CLASSES[i] for i in predictions.class_id.tolist()]

print(labels)

See LICENSE for the model license.

We love your input! Please see the core Autodistill contributing guide to get started. Thank you 🙏 to all our contributors!