Clean up

example.py

@@ -1,31 +1,32 @@
+# To run this example please download the PascalVOC 2007 dataset first:
+#
+# wget http://host.robots.ox.ac.uk/pascal/VOC/voc2007/VOCtrainval_06-Nov-2007.tar
+# tar -xvf VOCtrainval_06-Nov-2007.tar
+
 from pathlib import Path
 
-from labelformat.formats import YOLOv8ObjectDetectionInput
+from labelformat.formats import PascalVOCObjectDetectionInput
 
 from lightly_insights import analyze, present
 
-
-def main() -> None:
-    # Analyze an image folder.
-    image_folder = Path("/Users/michal/datasets/aquarium.v2-release.yolov8/test/images")
-    image_analysis = analyze.analyze_images(image_folder=image_folder)
-
-    # Analyze object detections.
-    input_file = Path("/Users/michal/datasets/aquarium.v2-release.yolov8/data.yaml")
-    label_input = YOLOv8ObjectDetectionInput(
-        input_file=input_file,
-        input_split="test",
-    )
-    od_analysis = analyze.analyze_object_detections(label_input=label_input)
-
-    # Create HTML report.
-    output_folder = Path("/Users/michal/tmp/lightly_insights_output")
-    present.create_html_report(
-        output_folder=output_folder,
-        image_analysis=image_analysis,
-        od_analysis=od_analysis,
-    )
-
-
-if __name__ == "__main__":
-    main()
+# Analyze an image folder.
+image_analysis = analyze.analyze_images(
+    image_folder=Path("./VOCdevkit/VOC2007/JPEGImages")
+)
+
+# Analyze object detections.
+label_input = PascalVOCObjectDetectionInput(
+    input_folder=Path("./VOCdevkit/VOC2007/Annotations"),
+    category_names=(
+        "person,bird,cat,cow,dog,horse,sheep,aeroplane,bicycle,boat,bus,car,"
+        "motorbike,train,bottle,chair,diningtable,pottedplant,sofa,tvmonitor"
+    ),
+)
+od_analysis = analyze.analyze_object_detections(label_input=label_input)
+
+# Create HTML report.
+present.create_html_report(
+    output_folder=Path("./html_report"),
+    image_analysis=image_analysis,
+    od_analysis=od_analysis,
+)

pyproject.toml

@@ -6,7 +6,7 @@ build-backend = "poetry.core.masonry.api"
 name = "lightly-insights"
 version = "0.1.0"
 authors = ["Lightly.ai"]
-description = "Easily get basic insights about your ML dataset."
+description = "Get quick insights about your ML dataset."
 readme = "README.md"
 license = "MIT"
 
@@ -29,9 +29,6 @@ build = "*"
 twine = "*"
 types-Pillow = "*"
 
-# [tool.poetry.scripts]
-# labelformat = "lightly_insights.cli.cli:main"
-
 [tool.pytest.ini_options]
 pythonpath = [
   ".", "src/"

src/lightly_insights/analyze.py

@@ -188,6 +188,7 @@ def analyze_object_detections(
             total_data.heatmap[int(y1) : int(y2), int(x1) : int(x2)] += 1
             class_datum.heatmap[int(y1) : int(y2), int(x1) : int(x2)] += 1
 
+            # Sample images.
             if len(class_datum.sample_filenames) < 4:
                 class_datum.sample_filenames.append(label.image.filename)
 

src/lightly_insights/plots.py

@@ -37,17 +37,6 @@ def create_object_plots(
     objects_per_image_path = plot_folder / "objects_per_image.png"
     heatmap_path = plot_folder / "heatmap.png"
 
-    # TODO: Remove.
-    if plot_folder.name != "plots":
-        return PlotPaths(
-            object_sizes_abs=str(object_sizes_abs_path.relative_to(output_folder)),
-            object_sizes_rel=str(object_sizes_rel_path.relative_to(output_folder)),
-            side_length_avg=str(side_length_avg_path.relative_to(output_folder)),
-            rel_area=str(rel_area_path.relative_to(output_folder)),
-            objects_per_image=str(objects_per_image_path.relative_to(output_folder)),
-            heatmap=str(heatmap_path.relative_to(output_folder)),
-        )
-
     # Bucket by multiples of 20px.
     size_histogram_abs = Counter(
         [
@@ -131,7 +120,7 @@ def create_object_plots(
 
 def width_heigth_pixels_plot(
     output_file: Path,
-    size_histogram: Counter[Tuple[float, float]],
+    size_histogram: Union[Counter[Tuple[float, float]], Counter[Tuple[int, int]]],
     title: str,
 ) -> None:
     # Image size plot.

src/lightly_insights/present.py

@@ -15,6 +15,7 @@
 
 logger = logging.getLogger(__name__)
 static_folder = Path(__file__).parent / "static"
+template_folder = Path(__file__).parent / "templates"
 
 
 @dataclass(frozen=True)
@@ -81,7 +82,7 @@ def create_html_report(
 
     # Setup Jinja2 environment
     env = Environment(
-        loader=FileSystemLoader(searchpath="./src/lightly_insights/templates"),
+        loader=FileSystemLoader(searchpath=template_folder),
         undefined=StrictUndefined,
     )
     template = env.get_template("report.html")
@@ -90,7 +91,7 @@ def create_html_report(
     html_output = template.render(report_data)
 
     # Write the HTML to file
-    html_output_path = output_folder / "report.html"
+    html_output_path = output_folder / "index.html"
     html_output_path.write_text(html_output)
 
     # Copy static files.
@@ -99,6 +100,8 @@ def create_html_report(
         shutil.rmtree(output_static_folder, ignore_errors=True)
     shutil.copytree(src=static_folder, dst=output_static_folder)
 
+    logger.info(f"Successfully created HTML report: {html_output_path}")
+
 
 def _get_image_insights(
     output_folder: Path,
@@ -107,14 +110,14 @@ def _get_image_insights(
     # Image size plot.
     plots.width_heigth_pixels_plot(
         output_file=output_folder / "image_size_plot.png",
-        size_histogram=image_analysis.image_sizes,  # type: ignore[arg-type]
+        size_histogram=image_analysis.image_sizes,
         title="Image Sizes",
     )
 
+    # Sample images.
     sample_folder = output_folder / "sample"
     sample_folder.mkdir(parents=True, exist_ok=True)
     sample_images = []
-    # TODO: Do this more efficiently.
     rng = random.Random(42)
     selection = rng.sample(sorted(list(image_analysis.filename_set)), k=8)
     for filename in selection:

tests/test_main.py

@@ -1,6 +1,3 @@
-from lightly_insights import main
-
-
-def test_main() -> None:
-    main.main()
+def test() -> None:
+    # Testing infrasctructure is set up!
     assert True