Example outputs:

• Painted image

---

API Reference

paint_ground_truth_detection(ground_truth_file, ground_truth_image_folder, output_folder, categories='categories', category_id='category_id', opacity=0.5)

Paint ground truth in detection format.

Parameters:

Name	Type	Description	Default
ground_truth_file	str	path to file in detection format.	required
ground_truth_image_folder	str	path to file of images associated with ground_truth_file.	required
output_folder	str	plotting results will be generated in this folder. If it does not exists, it will be created.	required
categories	str	name of the top level key holding information about the list of categories.	'categories'
category_id	str	name of the annotation level key holding information about the category index.	'category_id'
opacity	float	Set level of opacity of the annotations painted on top of the image. From 0 (transparent) to 1 (opaque).	0.5

Source code in src/stages/data/explore/paint_ground_truth_detection.py

def paint_ground_truth_detection(
    ground_truth_file: str,
    ground_truth_image_folder: str,
    output_folder: str,
    categories: str = "categories",
    category_id: str = "category_id",
    opacity: float = 0.5,
) -> None:
    """Paint ground truth in [detection](https://gradiant.github.io/ai-dataset-template/supported_tasks/#detection) format.

    Args:
        ground_truth_file:
            path to file in [detection](https://gradiant.github.io/ai-dataset-template/supported_tasks/#detection) format.
        ground_truth_image_folder:
            path to file of images associated with `ground_truth_file`.
        output_folder:
            plotting results will be generated in this folder.
            If it does not exists, it will be created.
        categories:
            name of the top level key holding information about the list of categories.
        category_id:
            name of the annotation level key holding information about the category index.
        opacity:
            Set level of opacity of the annotations painted on top of the image.
            From 0 (transparent) to 1 (opaque).
    """
    from PIL import Image
    from plotly import express as px
    from plotly import graph_objects as go

    ground_truth = json.load(open(ground_truth_file))

    category_id_to_name = {
        category["id"]: category["name"] for category in ground_truth[categories]
    }

    image_id_to_file_name = {
        image["id"]: image["file_name"] for image in ground_truth["images"]
    }

    image_id_to_annotations = defaultdict(list)
    for annotation in ground_truth["annotations"]:
        image_id_to_annotations[annotation["image_id"]].append(annotation)

    for image_id, annotation in image_id_to_annotations.items():
        file_name = image_id_to_file_name[image_id]

        logger.info(f"Loading {Path(ground_truth_image_folder) / file_name}")
        try:
            image_pil = Image.open(Path(ground_truth_image_folder) / file_name)
        except FileNotFoundError:
            logger.warning(f"{file_name} not found")
            continue

        fig = px.imshow(image_pil)
        fig.update_layout(title=Path(file_name).name)

        for annotation in image_id_to_annotations[image_id]:
            color = px.colors.qualitative.Light24[annotation[category_id] % 24]
            left, top, width, height = annotation["bbox"]
            right = left + width
            bottom = top + height
            fig.add_trace(
                go.Scatter(
                    x=[left, right, right, left],
                    y=[top, top, bottom, bottom],
                    fill="toself",
                    name=category_id_to_name[annotation[category_id]],
                    fillcolor=color,
                    line_color=color,
                    opacity=opacity,
                )
            )

        output_name = (Path(output_folder) / file_name).with_suffix(".html")
        output_name.parent.mkdir(parents=True, exist_ok=True)
        logger.info(f"Saving to {output_name}")
        fig.write_html(str(output_name))