[0.1.1] - 2025-07-07

This release significantly improves CSV parsing capabilities and robustness, making the application compatible with a wider variety of confusion matrix formats. Enhanced support for descriptive class labels and automatic class detection eliminates previous limitations and crashes, ensuring seamless analysis of real-world classification data.

Key Improvements

• Enhanced CSV Format Support

  • Added support for CSV files with descriptive ClassValue entries (e.g., C_1 - nezasazena uroda)
  • Automatic detection and processing of any number of classes (C_1, C_2, C_3, ...)
  • Robust parsing that extracts class codes from mixed descriptive text

• Stability and Error Handling

  • Fixed application crashes when ClassValue didn't match exact C_1, C_2 format
  • Improved error messages for invalid or empty CSV data
  • Enhanced validation for better user feedback

• Flexibility Enhancements

  • Metrics calculation now works seamlessly with various CSV formats
  • Removed restrictions on the number of classes that can be processed
  • More resilient data processing pipeline

Technical Improvements

• Strengthened CSV parsing logic to handle real-world data formats
• Enhanced error handling and user feedback mechanisms
• Improved compatibility with diverse confusion matrix export formats

---

Developed by:
Jakub Ešpandr – Born4Flight, FlyCamCzech
Licensed under the Non-Commercial Public License (NCPL v1.0)
Full Changelog: v0.1.0...v0.1.1

[0.1.0] - 2025-06-22

This release introduces the initial version of MetriCalc, a modern, localized application for evaluating classification performance using confusion matrices. With a clean UI, batch capabilities, and XLSX export, it simplifies the analysis of UAV, remote sensing, and image classification workflows.

New Features

• Modern User Interface

  • Built with PySide6 for a fast and responsive desktop experience
  • Clear tabbed layout for Single File and Batch Processing
  • Localized interface with real-time Czech/English switching

• Metric Calculation Engine

  • Parses semicolon-separated confusion matrix .csv files
  • Computes per-class and macro-averaged metrics:
    • Precision (User Accuracy)
    • Recall (Producer Accuracy)
    • F1-score
    • Overall Accuracy
    • Kappa Index

• Batch Processing Mode

  • Automatically process an entire folder of .csv files
  • Two output strategies:
    • One .xlsx per file
    • One consolidated .xlsx with a separate sheet for each result

• Excel Export

  • Localized headers based on selected UI language
  • Output includes two sheets:
    • Metrics: calculated scores
    • Data: original confusion matrix
  • File-safe sheet names and Unicode-safe formatting

• Performance Enhancements

  • Asynchronous threading ensures UI remains responsive
  • Progress dialog with cancel option for batch runs

Technical Details

• Fully object-oriented codebase split into core, ui, and utils
• Supports Python 3.8+
• Dependencies: PySide6, pandas, scikit-learn, openpyxl
  —
  Developed by:
  Jakub Ešpandr – Born4Flight, FlyCamCzech
  Licensed under the Non-Commercial Public License (NCPL v1.0)