Skip to content

SegmentationResult

spatialhub.structures.SegmentationResult represents spatial binary masks, bounding boxes, and confidence scores produced by segmentation, proposal, and zero-shot detection models such as FastSAMAdapter, SAMAdapter, and CNOSAdapter.

from spatialhub.structures import SegmentationResult

Fields

Field Type Shape Description
image np.ndarray (H, W, 3) uint8 Input RGB image.
boxes np.ndarray (N, 4) float32 Bounding boxes in [x1, y1, x2, y2] pixel coordinate format.
masks np.ndarray (N, H, W) bool Binary spatial masks, one per candidate detection.
scores np.ndarray (N,) float32 Detection or match confidence scores.
class_ids np.ndarray | None (N,) int Numeric class indices. None for class-agnostic proposals.
class_names list[str] | None Length N String class or CAD object labels.

Methods

visualize_mask

Renders colorized semi-transparent mask overlays, bounding box outlines, and confidence scores. Calls visualize_masks internally.

vis = result.visualize_mask(save_path="segmentation_overlay.png", alpha=0.4)

Parameters

Parameter Type Default Description
save_path str | Path | None None Optional disk path to write the annotated image.
alpha float 0.4 Mask blend transparency (\(0.0 \dots 1.0\)).

Return Value

  • np.ndarray: Annotated uint8 RGB image array of shape (H, W, 3).