Skip to content

Reference for ultralytics/models/fastsam/model.py

Note

This file is available at https://github.com/ultralytics/ultralytics/blob/main/ultralytics/models/fastsam/model.py. If you spot a problem please help fix it by contributing a Pull Request 🛠️. Thank you 🙏!


ultralytics.models.fastsam.model.FastSAM 

FastSAM(model='FastSAM-x.pt')

Bases: Model

FastSAM model interface for segment anything tasks.

This class extends the base Model class to provide specific functionality for the FastSAM (Fast Segment Anything Model) implementation, allowing for efficient and accurate image segmentation.

Attributes:

Name Type Description
model str

Path to the pre-trained FastSAM model file.
task str

The task type, set to "segment" for FastSAM models.

Examples:

>>> from ultralytics import FastSAM
>>> model = FastSAM("last.pt")
>>> results = model.predict("ultralytics/assets/bus.jpg")
Source code in ultralytics/models/fastsam/model.py 
28
29
30
31
32
33
def __init__(self, model="FastSAM-x.pt"):
    """Initialize the FastSAM model with the specified pre-trained weights."""
    if str(model) == "FastSAM.pt":
        model = "FastSAM-x.pt"
    assert Path(model).suffix not in {".yaml", ".yml"}, "FastSAM models only support pre-trained models."
    super().__init__(model=model, task="segment")

task_map property 

task_map

Returns a dictionary mapping segment task to corresponding predictor and validator classes.

predict 

predict(
    source,
    stream=False,
    bboxes=None,
    points=None,
    labels=None,
    texts=None,
    **kwargs
)

Perform segmentation prediction on image or video source.

Supports prompted segmentation with bounding boxes, points, labels, and texts. The method packages these prompts and passes them to the parent class predict method.

Parameters:

Name Type Description Default
source str | Image | ndarray

Input source for prediction, can be a file path, URL, PIL image, or numpy array.

 required
stream bool

Whether to enable real-time streaming mode for video inputs.

 False
bboxes list

Bounding box coordinates for prompted segmentation in format [[x1, y1, x2, y2], ...].

 None
points list

Point coordinates for prompted segmentation in format [[x, y], ...].

 None
labels list

Class labels for prompted segmentation.

 None
texts list

Text prompts for segmentation guidance.

 None
**kwargs Any

Additional keyword arguments passed to the predictor.

 {}

Returns:

Type Description
list

List of Results objects containing the prediction results.
Source code in ultralytics/models/fastsam/model.py 
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
def predict(self, source, stream=False, bboxes=None, points=None, labels=None, texts=None, **kwargs):
    """
    Perform segmentation prediction on image or video source.

    Supports prompted segmentation with bounding boxes, points, labels, and texts. The method packages these
    prompts and passes them to the parent class predict method.

    Args:
        source (str | PIL.Image | numpy.ndarray): Input source for prediction, can be a file path, URL, PIL image,
            or numpy array.
        stream (bool): Whether to enable real-time streaming mode for video inputs.
        bboxes (list): Bounding box coordinates for prompted segmentation in format [[x1, y1, x2, y2], ...].
        points (list): Point coordinates for prompted segmentation in format [[x, y], ...].
        labels (list): Class labels for prompted segmentation.
        texts (list): Text prompts for segmentation guidance.
        **kwargs (Any): Additional keyword arguments passed to the predictor.

    Returns:
        (list): List of Results objects containing the prediction results.
    """
    prompts = dict(bboxes=bboxes, points=points, labels=labels, texts=texts)
    return super().predict(source, stream, prompts=prompts, **kwargs)



📅 Created 1 year ago ✏️ Updated 6 months ago
jk4e glenn-jocher Burhan-Q