Skip to main content
Estimate metric depth from a single RGB image using Depth Anything 3. Returns depth values in meters.

​
Parameters

​
image_input
str | PIL.Image | np.ndarray
required
RGB image as file path, URL, PIL Image, or numpy array.
​
timeout
float | None
Optional timeout in seconds for the HTTP request.

​
Returns

np.ndarray — Metric depth map of shape (H, W) with dtype float32. Values represent depth in meters.

​
Example Output

DA3Metric input and depth map output

​
Example

from grid_cortex_client import CortexClient
from PIL import Image

client = CortexClient()
image = Image.open("scene.jpg")  # 640x480 RGB
depth = client.run(model_id="da3metric", image_input=image)

print(depth.shape, depth.dtype)
# (378, 504) float32 — note: output resolution differs from input

print(f"min={depth.min():.4f}, max={depth.max():.4f}, mean={depth.mean():.4f}")
# min=1.6177, max=53.1848, mean=12.9778  (meters)