Function annotate_image

tools/make_highlight.py:10–105

Draw a red rectangle highlighting the region and place a x2 zoomed inset of the region at the bottom-right corner of the image. Args: input_path: path to the input image. u, v: top-left corner of the highlight region (x=u, y=v). h, w: height and width of the hig

(
    input_path: str,
    u: int,
    v: int,
    h: int,
    w: int,
    out_path: str,
    thickness: int = 4,
    margin: int = 16,
    rect_color=(0, 0, 255),  # BGR: red
)

def annotate_image(
    input_path: str,
    u: int,
    v: int,
    h: int,
    w: int,
    out_path: str,
    thickness: int = 4,
    margin: int = 16,
    rect_color=(0, 0, 255),  # BGR: red
):
    """
    Draw a red rectangle highlighting the region and place a x2 zoomed inset
    of the region at the bottom-right corner of the image.

    Args:
        input_path: path to the input image.
        u, v: top-left corner of the highlight region (x=u, y=v).
        h, w: height and width of the highlight region.
        out_path: path to save the annotated output image.
        thickness: rectangle line thickness in pixels.
        margin: margin in pixels around the inset.
        rect_color: BGR color tuple for the rectangle (default red).
    """
    img = cv2.imread(input_path, cv2.IMREAD_UNCHANGED)
    if img is None:
        raise FileNotFoundError(f"Failed to read image: {input_path}")

    # If image has alpha channel, drop it for drawing operations.
    if img.ndim == 3 and img.shape[2] == 4:
        img = cv2.cvtColor(img, cv2.COLOR_BGRA2BGR)

    H, W = img.shape[:2]

    # Sanitize coordinates and clip to image bounds.
    x1 = int(max(0, u))
    y1 = int(max(0, v))
    x2 = int(min(W, u + w))
    y2 = int(min(H, v + h))

    if x2 <= x1 or y2 <= y1:
        raise ValueError("Highlight region is empty or out of bounds after clipping.")

    # Draw highlight rectangle.
    cv2.rectangle(img, (x1, y1), (x2, y2), rect_color, thickness=thickness)

    # Extract ROI and create x2 zoom (with fallback scaling if needed).
    roi = img[y1:y2, x1:x2]
    roi_h, roi_w = roi.shape[:2]

    target_w = roi_w * 2
    target_h = roi_h * 2

    # Compute max allowed size for inset (respecting margins).
    max_inset_w = max(1, W - 2 * margin)
    max_inset_h = max(1, H - 2 * margin)

    scale = min(