textify 0.5.4
textify: ^0.5.4 copied to clipboard
A cross-platform Flutter package that extracts text from clean digital images using standard fonts, run offline with no external dependencies.
Changelog #
All notable changes to this project will be documented in this file.
0.5.4 - 2026-05-01 #
Changed #
- Improved OCR accuracy for all test images (now 100% on all reference assets and generated images)
- Major refactor of post-processing pipeline for better modularity and maintainability
- Updated test suite and evaluation harness for stricter correctness and coverage
- Improved handling of punctuation, numbers, and mixed-case text
- Fixed edge cases in band splitting and artifact merging
- Updated English dictionary and normalization logic
0.5.3 (unpublished) #
Note: Version 0.5.3 was never published to pub.dev.
- Split tests into two runs "No English dictionary" and "with English dictionary"
0.5.2 : 2026-02-06 #
Changed #
- Quiet OCR eval output by default; keep verbose output when explicitly requested.
- Documentation and comment accuracy updates across the project.
0.5.1 : 2026-02-05 #
Changed #
- Replace magic numbers with named constants across core OCR components for clarity and maintainability.
- Added
fcheckto the verification process.
0.5.0 : 2025-12-27 #
Breaking Changes #
- Refactored Textify class to use TextifyConfig exclusively for configuration
- Removed deprecated setters from Textify class:
applyDictionary,excludeLongLines,dilatingSize, andinnerSplit. UseTextifyConfiginstead when creating Textify instances.
Updated #
- sdk: ^3.10.0
- flutter: ">=3.38.5"
- Use the GLAD tool to generate the Layer Diagram
0.4.6 : 2025-10-16 #
0.4.3 : 2025-04-07 #
Added #
- more API Documentation, PANA is now reporting 100% documentation coverage.
0.4.2 : 2025-04-06 #
0.4.1 : 2025-04-05 #
Changed #
- Improvement text matchings, now using average score of all templates fonts
- Code Coverage 99.3%
- Doc comments
0.4.0 : 2025-04-04 #
0.3.4 : 2025-01-14 #
0.3.1 : 2025-01-09 #
Changed #
- Updated README.md to show examples of good clean vs problematic text in input images
- Improved example using the sample of "Good Clean Image"
0.3.0 : 2025-01-07 #
0.2.0 : 2024-12-16 #
Changed #
- Adjusted band tolerance to 50% (was 80%)
- Adjusted word gap to 50% (was 75%)
- Adjusted gray scale to 190 (was 128)
0.1.5 : 2024-10-12 #
Changed #
- Updated Matrices.json to achieve 100% match for the 4 fonts at size 40
- Refactored and cleaned up code
0.1.3 : 2024-10-09 #
0.1.1 : 2024-10-06 #
Changed #
- Used embedded fonts for generating template [Arial, Courier, Helvetica, Times New Roman]
0.0.7 : 2024-10-01 #
0.0.6 : 2024-10-01 #
0.0.5 : 2024-09-30 #
0.0.4 : 2024-09-30 #
Added #
- Linked package to GitHub repo : https://github.com/vteam-com/textify