Overview

FastSAM is an image segmentation model trained using 2% of the data in the Segment Anything Model SA-1B dataset.

FastSAM overcomes the computation requirements barrier associated with using SAM by employing a decoupled approach.

FastSAM divided the segmentation task into two sequential stages: all-instance segmentation and prompt-guided selection.

‍

Use Cases:

Despite being exponentially smaller than its parent model, FastSAM still employs multiple use cases. The model is capable of not only image segmentation, but also zero-shot edge detection, image captioning, visual question answering, etc. Overall, FastSAM is a highly impressive model able to do a number of different computer tasks.

Performance

‍

While having a smaller memory usage, SAM is still able to perform well on a number of benchmarks.

‍

Use This Model

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


pip install autodistill-fastsam

Then, use this code to run inference:


from autodistill_fastsam import FastSAM
from autodistill.detection import CaptionOntology

# define an ontology to map class names to our FastSAM prompt
# the ontology dictionary has the format {caption: class}
# where caption is the prompt sent to the base model, and class is the label that will
# be saved for that caption in the generated annotations
# then, load the model
base_model = FastSAM(
    ontology=CaptionOntology(
        {
            "person": "person",
            "a forklift": "forklift"
        }
    )
)

results = base_model.predict("image.jpeg")

Label Data Automatically with FastSAM

You can automatically label a dataset using FastSAM with help from Autodistill, an open source package for training computer vision models. You can label a folder of images automatically with only a few lines of code. Below, see our tutorials that demonstrate how to use FastSAM to train a computer vision model.

FastSAM