Invisible Watermark

Common errors

• ModuleNotFoundError: No module named 'cv2'

  cause The `opencv-python` package, which provides the `cv2` module, is either not installed or not correctly installed in your environment.
  fix Install `opencv-python` using pip: `pip install opencv-python`. If you are in a headless environment, try `pip install opencv-python-headless`.

• AttributeError: 'NoneType' object has no attribute 'shape'

  cause `cv2.imread()` returned `None` because the specified image file could not be found, is corrupted, or is in an unsupported format. The library then attempts to access attributes of this `None` object.
  fix Verify that the image path is correct and the file exists. Check if the image file is valid and readable by OpenCV by trying to load it directly with `cv2.imread()` in a Python interpreter.

• IndexError: too many indices for array

  cause This typically occurs when the library expects a multi-channel image (e.g., BGR) but receives a single-channel grayscale image or an image with unexpected dimensions.
  fix Ensure that the input image is a color image. If you have a grayscale image, convert it to a 3-channel image before passing it to `iwm.encode()`, for example: `import cv2; gray_img = cv2.imread('gray.png', cv2.IMREAD_GRAYSCALE); color_img = cv2.cvtColor(gray_img, cv2.COLOR_GRAY2BGR)`.

Warnings

• gotcha The library heavily relies on `opencv-python` for image processing. Installation of `opencv-python` can sometimes be challenging or lead to `ModuleNotFoundError: No module named 'cv2'` on certain systems (e.g., ARM processors, headless servers without GUI dependencies).
  Fix: Ensure `opencv-python` (or `opencv-python-headless` for server environments) is successfully installed. You may need to manually verify `import cv2` works. For persistent issues, consult OpenCV installation guides for your specific OS/architecture.
• gotcha The `encode` method expects image paths that `cv2.imread()` can successfully load. Using incorrect file paths, unsupported image formats, or corrupted images will result in failures or `NoneType` errors.
  Fix: Always double-check image paths for correctness (absolute vs. relative). Prior to using the library, verify images are readable by OpenCV: `import cv2; img = cv2.imread('your_image.jpg'); if img is None: print('Failed to load image')`.
• gotcha As an early-stage library (version 0.2.0), the API is still evolving. While specific breaking changes weren't extensively documented between 0.1.x and 0.2.x, future minor or patch versions could introduce them without comprehensive migration guides.
  Fix: It is recommended to pin your `invisible-watermark` version in your `requirements.txt` (e.g., `invisible-watermark==0.2.0`) and review release notes or the GitHub repository for any breaking changes before upgrading.

Install

• pip install invisible-watermark Install stable version

Imports

• InvisibleWatermark

  from invisible_watermark import InvisibleWatermark

Quickstart

This quickstart demonstrates how to create a dummy image, then initialize the InvisibleWatermark library to encode a text watermark into it, and finally decode the watermark from the saved output image.

import os
from PIL import Image
from invisible_watermark import InvisibleWatermark

# 1. Create a dummy image for demonstration if it doesn't exist
image_filename = "dummy_input_image.png"
output_filename = "dummy_watermarked_image.png"

if not os.path.exists(image_filename):
    img = Image.new('RGB', (256, 256), color = 'red')
    img.save(image_filename)
    print(f"Created dummy image: {image_filename}")

# 2. Initialize the watermark library
iwm = InvisibleWatermark()

# 3. Define the watermark text
watermark_text = "SecretMessage123"
print(f"Original watermark: {watermark_text}")

# 4. Encode the watermark into the image
iwm.encode(image_filename, output_filename, watermark_text)
print(f"Watermark encoded and saved to: {output_filename}")

# 5. Decode the watermark from the watermarked image
decoded_text = iwm.decode(output_filename)
print(f"Decoded watermark: {decoded_text}")

# 6. Clean up generated files (optional)
# os.remove(image_filename)
# os.remove(output_filename)