Depth-Anything-V2

This example demonstrates using the Depth-Anything-V2 tool for depth estimation on images.

from vision_agent_tools.models.depth_anything_v2 import DepthAnythingV2

# (replace this path with your own!)
test_image = "path/to/your/image.jpg"

# Load the image
image = Image.open(test_image)
# Initialize the depth map estimation model.
depth_estimate = DepthAnythingV2()

# Run the inference
results = depth_estimate(image)

# Let's print the obtained depth map
print(results.map)

DepthAnythingV2

Bases: BaseMLModel

Model for depth estimation using the Depth-Anything-V2 model from the paper Depth Anything V2.

__call__(image, grayscale=False)

Depth-Anything-V2 is a highly practical solution for robust monocular depth estimation.

Parameters:

Name	Type	Description	Default
image	Union[str, Image, ndarray]	The input image for depth estimation. Can be a file path, a PIL Image, or a NumPy array.	required
grayscale	bool	Whether to return the depth map as a grayscale image. If True, the depth map will be normalized to the range [0, 255] and converted to uint8. Defaults to False.	False

Returns:

Name	Type	Description
DepthMap	DepthMap	An object type containing a numpy array with the HxW depth map of the image.

__init__()

Initializes the Depth-Anything-V2 model.

DepthMap

Bases: BaseModel

Represents the depth map of an image.

Attributes:

Name	Type	Description
map	Any	HxW raw depth map of the image.