-s, --submissions Required. Define the submissions' path. It must contain multiple folders (for each submission).
-g, --gt Required. The Ground Truth Image Files.
-w, --weights Required. The Path of the weights' directory of the ground truth image (required for evaluation) with the same name with the GT images.
-o, --outputfile Required. Write the results to a CSV compatible file.
--detailedoutput (Default: false) Produce a detailed scored for each metric for each submission for each ground-truth image.
--perimagescore (Default: true) Calculate the final score based on image position (DIBCO Competitions) instead of position on averages (H-DIBCO 2018)
--help Display this help screen.
--version Display version information.
- Verify the MATLAB Compiler Runtime (MCR) is installed and ensure you have installed version MRC 9.4. You may download it from from https://www.mathworks.com/products/compiler/matlab-runtime.html
The executable BinEvalWeights.exe generates two .dat files containing the normalized distance weights for pseudo-Recall/Precision based on paper 1.
Input: Filename of the binary ground-truth image (supported formats of OpenCV ver.1).
Output: __RWeights.dat and __PWeights.dat
Filenames of:
- GT image
- Binarized image for evaluation
- "Recall Weights" .dat file *
- "Precision Weights" .dat file *
- F-Measure
- pseudo F-Measure (Fps)*
- PSNR
- DRD
The outputs are CSV ';' delimited. For example (see TestResources/
):
.\DIBCO_metrics.exe .\5_gt.bmp .\5.bmp .\5_RWeights.dat .\5_PWeights.dat
produces the following output:
F-Measure;pseudo F-Measure;PSNR;DRD
66.214;71.4372;14.5629;13.8647
*Notice that for I.3 and I.4 a different executable program is required to generate the .dat files containing the "Recall/Precision weights". It is provided at Prerequisites/BinEvalWeights
or at http://users.iit.demokritos.gr/~kntir/TIP_exe/.**
DIBCO_metrics.exe .\5_gt.bmp .\5.bmp .\5_RWeights.dat .\5_PWeights.dat
CAUTION: When providing the .dat files, always provide the "Recall weights" .dat file first and afterwards the "Precision weights" .dat file.
-
K. Ntirogiannis, B. Gatos and I. Pratikakis, "Performance Evaluation Methodology for Historical Document Image Binarization", IEEE Trans. Image Proc., vol.22, no.2, pp. 595-609, Feb. 2013.
-
H.Lu, A.C. Kot and Y.Q. Shi, "Distance Reciprocal Distortion Measure for Binary Document Images", IEEE Sigal Proc. Lett., vol.11, no.2, pp. 228-231, Feb. 2004.