Add "Get Bounding Box" node

backend/src/packages/chaiNNer_standard/image_dimension/utility/get_bounding_box.py

@@ -0,0 +1,53 @@
+from __future__ import annotations
+
+from typing import Tuple
+
+import numpy as np
+
+from nodes.properties.inputs import ImageInput, SliderInput
+from nodes.properties.outputs import NumberOutput
+from nodes.utils.utils import get_h_w_c
+
+from .. import utility_group
+
+
+@utility_group.register(
+    schema_id="chainner:image:get_bbox",
+    name="Get Bounding Box",
+    description="Gets a bounding box (X, Y, Height, and Width) of the white area of a mask.",
+    icon="BsBoundingBox",
+    inputs=[
+        ImageInput(channels=1),
+        SliderInput(
+            "Threshold",
+            precision=1,
+            minimum=0,
+            maximum=100,
+            controls_step=1,
+            slider_step=1,
+            default=0,
+        ),
+    ],
+    outputs=[
+        NumberOutput("X", output_type="min(uint, Input0.width - 1)"),
+        NumberOutput("Y", output_type="min(uint, Input0.height - 1)"),
+        NumberOutput("Width", output_type="min(uint, Input0.width) & 1.."),
+        NumberOutput("Height", output_type="min(uint, Input0.height) & 1.."),
+    ],
+)
+def get_bounding_box_node(
+    img: np.ndarray,
+    thresh_val: float,
+) -> Tuple[int, int, int, int]:
+    # Threshold value 100 guarantees an empty image, so make sure the max
+    # is just below that.
+    thresh = min(thresh_val / 100, 0.99999)
+    h, w, _ = get_h_w_c(img)
+
+    r = np.any(img > thresh, 1)
+    c = np.any(img > thresh, 0)
+    if not r.any():
+        raise RuntimeError("Resulting bounding box is empty.")
+
+    x, y = c.argmax(), r.argmax()
+    return int(x), int(y), int(w - x - c[::-1].argmax()), int(h - y - r[::-1].argmax())