Skip to content

Correction AnalyzerΒΆ

IntroductionΒΆ

This package provides a class for analyzing color correction by combining various correction methods with object detection models. The analyzer offers multiple correction methods and customizable parameters that you can experiment with. For object detection, it currently supports YOLOv8 (ONNX), with plans to expand to more models in the future. For correction methods, it supports least squares, linear regression, affine regression, and polynomial regression.

The analyzer automatically runs all defined methods and generates a comprehensive report comparing visual results and Ξ”E CIE 2000 values. You don't need to write complex code - simply define your desired methods and parameters, and let the analyzer do the work for you. This makes it easy to find the best color correction approach for your specific needs.

UsageΒΆ

If you don't have image to test

You can download the sample image from the following link: 

curl -L -o input_image.jpg "https://drive.google.com/uc?export=download&id=1syOqw9kC0tt01p7yEobU4MeLfh336DZA"


# your_script.py
import cv2

from color_correction import ColorCorrectionAnalyzer

input_image_path = "your_path_image"

report = ColorCorrectionAnalyzer(
    list_correction_methods=[
        ("least_squares", {}),
        ("linear_reg", {}),
        ("affine_reg", {}),
        ("polynomial", {"degree": 2}),
        ("polynomial", {"degree": 3}),
        ("polynomial", {"degree": 4}),
        ("polynomial", {"degree": 5}),
    ],
    list_detection_methods=[
        ("yolov8", {"detection_conf_th": 0.25}),
    ],
    use_gpu=False,
)

df_report = report.run(
    input_image=cv2.imread(input_image_path),
    reference_image=None,
    output_dir="report-output", # (1)
)

df_report.head()
  1. πŸ’¬ The output directory where the report files will be saved.
β”œβ”€β”€ πŸ“„ your_script.py
└── πŸ“‚ report-output # (1)
    β”œβ”€β”€ report.html
    β”œβ”€β”€ report.pkl
    └── report_no_image.csv
  1. πŸ’¬ The output directory contains the following files:
    • report.html: The HTML report file.
    • report.pkl: The pickle file containing the report data.
    • report_no_image.csv: The CSV file containing the report data without images.

Report HTML Output

Comparison of Color Correction Methods: Visual Results and Ξ”E CIE 2000 Values (Click to enlarge)

ReferenceΒΆ