Function _plot_counting

vision_agent/tools/tools.py:3490–3541

(
    image: Image.Image,
    bboxes: List[Dict[str, Any]],
    colors: Dict[str, Tuple[int, int, int]],
    use_tracking_label: bool = False,
)

def _plot_counting(
    image: Image.Image,
    bboxes: List[Dict[str, Any]],
    colors: Dict[str, Tuple[int, int, int]],
    use_tracking_label: bool = False,
) -> Image.Image:
    width, height = image.size
    fontsize = max(12, int(min(width, height) / 40))
    draw = ImageDraw.Draw(image)
    font = ImageFont.truetype(
        str(resources.files("vision_agent.fonts").joinpath("default_font_ch_en.ttf")),
        fontsize,
    )
    for i, elt in enumerate(bboxes, 1):
        if use_tracking_label:
            label = elt["label"].split(":")[0]
            color = colors[elt["label"].split(":")[1].strip()]
        else:
            label = f"{i}"
            color = colors[elt["label"]]
        box = elt["bbox"]

        # denormalize the box if it is normalized
        box = denormalize_bbox(box, (height, width))
        x0, y0, x1, y1 = box
        cx, cy = (x0 + x1) / 2, (y0 + y1) / 2

        text_box = draw.textbbox(
            (cx, cy), text=label, font=font, align="center", anchor="mm"
        )

        # Calculate the offset to center the text within the bounding box
        text_width = text_box[2] - text_box[0]
        text_height = text_box[3] - text_box[1]
        text_x0 = cx - text_width / 2
        text_y0 = cy - text_height / 2
        text_x1 = cx + text_width / 2
        text_y1 = cy + text_height / 2

        # Draw the rectangle encapsulating the text
        draw.rectangle((text_x0, text_y0, text_x1, text_y1), fill=color)

        # Draw the text at the center of the bounding box
        draw.text(
            (text_x0, text_y0),
            label,
            fill="black",
            font=font,
            anchor="lt",
        )

    return image


FUNCTION_TOOLS = [