protect_image()

Detect and encrypt all license plates found in a PIL image.

Signature

safelicensing.protect_image(
    image: PIL.Image.Image,
    seed: float = 0.5,
    model = None,
    model_path: str | None = None,
) -> ProtectImageResult

Parameters

Parameter	Type	Default	Description
image	PIL.Image.Image	required	Input image. Any colour mode accepted; converted to RGB internally.
seed	float	0.5	Chaotic encryption seed in (0.0, 1.0) exclusive. Same seed on same image always produces same encrypted output.
model	YOLO instance	None	Pre-loaded YOLO model. Loads bundled best.pt when None.
model_path	str	None	Path to custom YOLOv8 .pt weights. Ignored when model is given.

Returns: ProtectImageResult

Field	Type	Description
original	PIL.Image	Unmodified RGB input
detected	PIL.Image	Copy of input with red bounding boxes drawn around detected plates
encrypted	PIL.Image	Full image with every detected plate region encrypted
bboxes	list[tuple[int, int, int, int]]	(x1, y1, x2, y2) bounding box per detected plate
elapsed	float	Wall-clock processing time in seconds

Raises

Exception	Condition
TypeError	image is not a PIL.Image.Image instance
ValueError	seed is not strictly inside (0.0, 1.0)

Examples

Basic usage

import safelicensing as sl
from PIL import Image

model = sl.load_model()
result = sl.protect_image(Image.open("car.jpg"), seed=0.42, model=model)

result.encrypted.save("car_protected.jpg")
print(result.bboxes)    # [(x1, y1, x2, y2), ...]
print(result.elapsed)   # 0.35

Load model once, process many images

import safelicensing as sl
from pathlib import Path
from PIL import Image

model = sl.load_model()

for path in Path("dataset/").glob("*.jpg"):
    result = sl.protect_image(Image.open(path), seed=0.5, model=model)
    result.encrypted.save(f"output/{path.name}")

Custom model path

model = sl.load_model("my_custom_weights.pt")
result = sl.protect_image(image, seed=0.3, model=model)

Access all output fields

result = sl.protect_image(image, seed=0.7)

print(f"Plates detected : {len(result.bboxes)}")
print(f"Bounding boxes  : {result.bboxes}")
print(f"Elapsed         : {result.elapsed:.3f}s")

result.original.save("original.jpg")
result.detected.save("detected.jpg")   # red boxes
result.encrypted.save("encrypted.jpg")

Validate the encryption seed

try:
    result = sl.protect_image(image, seed=1.0)   # raises ValueError
except ValueError as e:
    print(e)  # seed must be in (0.0, 1.0) exclusive, got 1.0

Notes

• The original image is never mutated; all outputs are independent copies.
• When no plates are detected, encrypted is identical to original and bboxes is empty.
• The elapsed timer starts after model loading. It measures detection + encryption only.

---

See also: protect_video() → | load_model() →