feat: adjust visualization to scale with image size (#482)

* GETI-480-adjust-visualization-to-image-size

* copilot comments

Author: mgumowsk

src/model_api/visualizer/defaults.py

@@ -0,0 +1,25 @@
+"""Default visualization constants."""
+
+# Copyright (C) 2026 Intel Corporation
+# SPDX-License-Identifier: Apache-2.0
+
+# Font sizes
+DEFAULT_FONT_SIZE: int = 16
+"""Default font size used for all text (labels, bounding boxes, overlays, keypoints)."""
+
+# Line / outline widths
+DEFAULT_OUTLINE_WIDTH: int = 2
+"""Default outline width for bounding boxes and polygon contours."""
+
+# Opacity
+DEFAULT_OPACITY: float = 0.4
+"""Default blend opacity for overlays and polygon fills."""
+
+# Keypoint drawing
+DEFAULT_KEYPOINT_SIZE: int = 3
+"""Default radius (in pixels) for keypoint dots."""
+
+# Scale baseline
+SCALE_BASELINE: int = 1280
+"""Longer-edge pixel count of 720p (landscape). Used as the denominator when
+computing the auto-scale factor."""

src/model_api/visualizer/layout/hstack.py

@@ -34,7 +34,11 @@ def _compute_on_primitive(self, primitive: Type[Primitive], image: PIL.Image, sc
             for _primitive in scene.get_primitives(primitive):
                 image_ = _primitive.compute(image.copy())
                 if isinstance(_primitive, Overlay):
-                    image_ = Overlay.overlay_labels(image=image_, labels=_primitive.label)
+                    image_ = Overlay.overlay_labels(
+                        image=image_,
+                        labels=_primitive.label,
+                        font_size=_primitive.font_size,
+                    )
                 images.append(image_)
             return self._stitch(*images)
         return None

src/model_api/visualizer/primitive/bounding_box.py

@@ -5,7 +5,9 @@
 
 from __future__ import annotations
 
-from PIL import Image, ImageDraw
+from PIL import Image, ImageDraw, ImageFont
+
+from model_api.visualizer.defaults import DEFAULT_FONT_SIZE, DEFAULT_OUTLINE_WIDTH
 
 from .primitive import Primitive
 
@@ -20,6 +22,8 @@ class BoundingBox(Primitive):
         y2 (int): y-coordinate of the bottom-right corner of the bounding box.
         label (str | None): Label of the bounding box.
         color (str | tuple[int, int, int]): Color of the bounding box.
+        outline_width (int): Width of the bounding box outline.
+        font_size (int): Font size for the label text.
 
     Example:
         >>> bounding_box = BoundingBox(x1=10, y1=10, x2=100, y2=100, label="Label Name")
@@ -34,30 +38,35 @@ def __init__(
         y2: int,
         label: str | None = None,
         color: str | tuple[int, int, int] = "blue",
+        outline_width: int = DEFAULT_OUTLINE_WIDTH,
+        font_size: int = DEFAULT_FONT_SIZE,
     ) -> None:
         self.x1 = x1
         self.y1 = y1
         self.x2 = x2
         self.y2 = y2
         self.label = label
         self.color = color
-        self.y_buffer = 5  # Text at the bottom of the text box is clipped. This prevents that.
+        self.outline_width = outline_width
+        self.font_size = font_size
+        self.font = ImageFont.load_default(size=self.font_size)
+        self.y_buffer = max(3, font_size // 3)  # Text at the bottom of the text box is clipped. This prevents that.
 
     def compute(self, image: Image) -> Image:
         draw = ImageDraw.Draw(image)
         # draw rectangle
-        draw.rectangle((self.x1, self.y1, self.x2, self.y2), outline=self.color, width=2)
+        draw.rectangle((self.x1, self.y1, self.x2, self.y2), outline=self.color, width=self.outline_width)
         # add label
         if self.label:
             # draw the background of the label
-            textbox = draw.textbbox((0, 0), self.label)
+            textbox = draw.textbbox((0, 0), self.label, font=self.font)
             label_image = Image.new(
                 "RGB",
                 (textbox[2] - textbox[0], textbox[3] + self.y_buffer - textbox[1]),
                 self.color,
             )
             draw = ImageDraw.Draw(label_image)
             # write the label on the background
-            draw.text((0, 0), self.label, fill="white")
+            draw.text((0, 0), self.label, font=self.font, fill="white")
             image.paste(label_image, (self.x1, self.y1))
         return image

src/model_api/visualizer/primitive/keypoints.py

@@ -8,6 +8,8 @@
 import numpy as np
 from PIL import Image, ImageDraw, ImageFont
 
+from model_api.visualizer.defaults import DEFAULT_FONT_SIZE, DEFAULT_KEYPOINT_SIZE
+
 from .primitive import Primitive
 
 
@@ -25,12 +27,14 @@ def __init__(
         keypoints: np.ndarray,
         scores: Union[np.ndarray, None] = None,
         color: Union[str, tuple[int, int, int]] = "purple",
-        keypoint_size: int = 3,
+        keypoint_size: int = DEFAULT_KEYPOINT_SIZE,
+        font_size: int = DEFAULT_FONT_SIZE,
     ) -> None:
         self.keypoints = self._validate_keypoints(keypoints)
         self.scores = scores
         self.color = color
         self.keypoint_size = keypoint_size
+        self.font_size = font_size
 
     def compute(self, image: Image) -> Image:
         """Draw keypoints on the image."""
@@ -47,7 +51,7 @@ def compute(self, image: Image) -> Image:
             )
 
         if self.scores is not None:
-            font = ImageFont.load_default(size=18)
+            font = ImageFont.load_default(size=self.font_size)
             for score, keypoint in zip(self.scores, self.keypoints):
                 textbox = draw.textbbox((0, 0), f"{score:.2f}", font=font)
                 draw.text(

src/model_api/visualizer/primitive/label.py

@@ -8,6 +8,8 @@
 
 from PIL import Image, ImageDraw, ImageFont
 
+from model_api.visualizer.defaults import DEFAULT_FONT_SIZE
+
 from .primitive import Primitive
 
 
@@ -46,7 +48,7 @@ def __init__(
         fg_color: Union[str, tuple[int, int, int]] = "black",
         bg_color: Union[str, tuple[int, int, int]] = "yellow",
         font_path: Union[str, BytesIO, None] = None,
-        size: int = 16,
+        size: int = DEFAULT_FONT_SIZE,
     ) -> None:
         self.label = f"{label} ({score:.2f})" if score is not None else label
         self.fg_color = fg_color

src/model_api/visualizer/primitive/overlay.py

@@ -11,6 +11,8 @@
 import PIL
 from PIL import ImageFont
 
+from model_api.visualizer.defaults import DEFAULT_FONT_SIZE, DEFAULT_OPACITY
+
 from .primitive import Primitive
 
 
@@ -28,12 +30,14 @@ class Overlay(Primitive):
     def __init__(
         self,
         image: PIL.Image | np.ndarray,
-        opacity: float = 0.4,
+        opacity: float = DEFAULT_OPACITY,
         label: Union[str, None] = None,
+        font_size: int = DEFAULT_FONT_SIZE,
     ) -> None:
         self.image = self._to_pil(image)
         self.label = label
         self.opacity = opacity
+        self.font_size = font_size
 
     def _to_pil(self, image: PIL.Image | np.ndarray) -> PIL.Image:
         if isinstance(image, np.ndarray):
@@ -45,15 +49,25 @@ def compute(self, image: PIL.Image) -> PIL.Image:
         return PIL.Image.blend(image, image_, self.opacity)
 
     @classmethod
-    def overlay_labels(cls, image: PIL.Image, labels: Union[list[str], str, None] = None) -> PIL.Image:
+    def overlay_labels(
+        cls,
+        image: PIL.Image,
+        labels: Union[list[str], str, None] = None,
+        font_size: int = DEFAULT_FONT_SIZE,
+    ) -> PIL.Image:
         """Draw labels at the bottom center of the image.
 
         This is handy when you want to add a label to the image.
+
+        Args:
+            image: Image to overlay the labels on.
+            labels: Labels to overlay.
+            font_size: Font size for the label text.
         """
         if labels is not None:
             labels = [labels] if isinstance(labels, str) else labels
-            font = ImageFont.load_default(size=18)
-            buffer_y = 5
+            font = ImageFont.load_default(size=font_size)
+            buffer_y = max(3, font_size // 3)
             dummy_image = PIL.Image.new("RGB", (1, 1))
             draw = PIL.ImageDraw.Draw(dummy_image)
             textbox = draw.textbbox((0, 0), ", ".join(labels), font=font)

src/model_api/visualizer/primitive/polygon.py

@@ -11,6 +11,8 @@
 import cv2
 from PIL import Image, ImageColor, ImageDraw
 
+from model_api.visualizer.defaults import DEFAULT_OPACITY, DEFAULT_OUTLINE_WIDTH
+
 from .primitive import Primitive
 
 if TYPE_CHECKING:
@@ -41,8 +43,8 @@ def __init__(
         points: list[tuple[int, int]] | None = None,
         mask: np.ndarray | None = None,
         color: str | tuple[int, int, int] = "blue",
-        opacity: float = 0.4,
-        outline_width: int = 2,
+        opacity: float = DEFAULT_OPACITY,
+        outline_width: int = DEFAULT_OUTLINE_WIDTH,
     ) -> None:
         self.points = self._get_points(points, mask)
         self.color = color

src/model_api/visualizer/scene/anomaly.py

@@ -3,13 +3,13 @@
 # Copyright (C) 2024 Intel Corporation
 # SPDX-License-Identifier: Apache-2.0
 
-from itertools import starmap
 from typing import Union
 
 import cv2
 from PIL import Image
 
 from model_api.models.result import AnomalyResult
+from model_api.visualizer.defaults import DEFAULT_FONT_SIZE, DEFAULT_OUTLINE_WIDTH
 from model_api.visualizer.layout import Flatten, Layout
 from model_api.visualizer.primitive import BoundingBox, Label, Overlay, Polygon
 
@@ -19,7 +19,14 @@
 class AnomalyScene(Scene):
     """Anomaly Scene."""
 
-    def __init__(self, image: Image, result: AnomalyResult, layout: Union[Layout, None] = None) -> None:
+    def __init__(
+        self,
+        image: Image,
+        result: AnomalyResult,
+        layout: Union[Layout, None] = None,
+        scale: float = 1.0,
+    ) -> None:
+        self.scale = scale
         super().__init__(
             base=image,
             overlay=self._get_overlays(result),
@@ -32,23 +39,44 @@ def __init__(self, image: Image, result: AnomalyResult, layout: Union[Layout, No
     def _get_overlays(self, result: AnomalyResult) -> list[Overlay]:
         if result.anomaly_map is not None:
             anomaly_map = cv2.cvtColor(result.anomaly_map, cv2.COLOR_BGR2RGB)
-            return [Overlay(anomaly_map)]
+            return [Overlay(anomaly_map, font_size=int(DEFAULT_FONT_SIZE * self.scale))]
         return []
 
     def _get_bounding_boxes(self, result: AnomalyResult) -> list[BoundingBox]:
         if result.pred_boxes is not None:
-            return list(starmap(BoundingBox, result.pred_boxes))
+            return [
+                BoundingBox(
+                    x1=box[0],
+                    y1=box[1],
+                    x2=box[2],
+                    y2=box[3],
+                    outline_width=max(1, int(DEFAULT_OUTLINE_WIDTH * self.scale)),
+                    font_size=int(DEFAULT_FONT_SIZE * self.scale),
+                )
+                for box in result.pred_boxes
+            ]
         return []
 
     def _get_labels(self, result: AnomalyResult) -> list[Label]:
         labels = []
         if result.pred_label is not None and result.pred_score is not None:
-            labels.append(Label(label=result.pred_label, score=result.pred_score))
+            labels.append(
+                Label(
+                    label=result.pred_label,
+                    score=result.pred_score,
+                    size=int(DEFAULT_FONT_SIZE * self.scale),
+                ),
+            )
         return labels
 
     def _get_polygons(self, result: AnomalyResult) -> list[Polygon]:
         if result.pred_mask is not None:
-            return [Polygon(result.pred_mask)]
+            return [
+                Polygon(
+                    result.pred_mask,
+                    outline_width=max(1, int(DEFAULT_OUTLINE_WIDTH * self.scale)),
+                ),
+            ]
         return []
 
     @property

src/model_api/visualizer/scene/classification.py

@@ -9,6 +9,7 @@
 from PIL import Image
 
 from model_api.models.result import ClassificationResult
+from model_api.visualizer.defaults import DEFAULT_FONT_SIZE
 from model_api.visualizer.layout import Flatten, Layout
 from model_api.visualizer.primitive import Label, Overlay
 
@@ -18,7 +19,14 @@
 class ClassificationScene(Scene):
     """Classification Scene."""
 
-    def __init__(self, image: Image, result: ClassificationResult, layout: Union[Layout, None] = None) -> None:
+    def __init__(
+        self,
+        image: Image,
+        result: ClassificationResult,
+        layout: Union[Layout, None] = None,
+        scale: float = 1.0,
+    ) -> None:
+        self.scale = scale
         super().__init__(
             base=image,
             label=self._get_labels(result),
@@ -31,7 +39,13 @@ def _get_labels(self, result: ClassificationResult) -> list[Label]:
         if result.top_labels is not None and len(result.top_labels) > 0:
             for label in result.top_labels:
                 if label.name is not None:
-                    labels.append(Label(label=label.name, score=label.confidence))
+                    labels.append(
+                        Label(
+                            label=label.name,
+                            score=label.confidence,
+                            size=int(DEFAULT_FONT_SIZE * self.scale),
+                        ),
+                    )
         return labels
 
     def _get_overlays(self, result: ClassificationResult) -> list[Overlay]: