Conversation
- Add input validation before assigning to self.value - Validate that input is a string and not empty - Only set self.value after successful directory creation and validation - Addresses requirement to validate all values before they get set
- Created .github/workflows/pytest-coverage.yml for automated testing - Tests run on Python 3.9, 3.10, 3.11, 3.12 for every PR - Added pytest-cov to setup.py dependencies - Coverage reports uploaded to QLTY Cloud (free for open source) - Uses OIDC authentication following QLTY best practices - Current coverage: 43% (48/48 tests passing)
- Created QRCodeDetector utility for extracting variety info from QR codes - Integrated QR detection into Segmentation analysis - QR detection is optional - continues normally if no QR code found - Saves variety info (variety_code, timing, full_variety) to variety_info.txt - Expected QR format: 'BB-Late', 'CC-Early', etc. - Tested: Works with and without QR codes in images
- Interactive script to generate QR codes with experimental metadata - Fields: project code, lot code, date, variety - Format: PROJECT|LOT|DATE|VARIETY (pipe-delimited) - Outputs to qr_codes/ directory with descriptive filenames
- Moved hardcoded starch scale data from StarchScales class to external YAML file - Added load_starch_scales() function to read Granny/config/starch_scales.yml - Dynamically populate StarchScales class attributes from YAML at module load time - Reduced StarchArea.py by 102 lines by externalizing data - Improves maintainability: starch scales can now be updated without code changes
Refactor starch scales to load from YAML configuration
- Moved hardcoded starch scale data from StarchScales class to external YAML file - Added load_starch_scales() function to read Granny/assets/starch_scales.yml - Dynamically populate StarchScales class attributes from YAML at module load time - Reduced StarchArea.py by 102 lines by externalizing data - Improves maintainability: starch scales can now be updated without code changes
Refactor starch scales to load from YAML asset file
Integrate QR code detection throughout the analysis pipeline to automatically capture and propagate experimental metadata (project, lot, date, variety) from tray labels to analysis outputs. Changes: - Update QRCodeDetector to parse pipe-delimited QR format (PROJECT|LOT|DATE|VARIETY) - Modify Segmentation to include QR data in segmented image filenames when detected - Add helper method to Analysis base class for extracting QR data from filenames - Update all downstream analyses (Starch, Blush, PeelColor, SuperficialScald) to extract and include QR metadata in CSV outputs - Fix MetaDataValue to use numeric_only=True when calculating tray averages to handle string metadata columns - Update QR code generator to output to QR_Output directory When QR code is detected, segmented images are named: PROJECT_LOT_DATE_VARIETY_fruit_##.png When no QR code is detected, default naming is used: original_tray_name_fruit_##.png CSV outputs now include project, lot, date, and variety columns populated from QR data when available, otherwise left empty for backward compatibility.
The QR code generator is a utility tool for creating tray labels and does not need to be part of the core Granny analysis package. Users can create QR codes using standard QR generation tools or libraries as needed for their workflows.
- Added new starch scale calibrations: PURDUE, DANJOU, AMBROSIA, MINNEISKA, PINKLADY, REDDELICIOUS1980, REDDELICIOUS1990, ROYAL_GALA - Implemented configurable RGB text color parameters for segmentation visualization (text_color_r, text_color_g, text_color_b) - All new varieties processed with default threshold 172 - Total starch scales now: 17 varieties
- Changed PURDUE calibration to use default threshold 172 - Added threshold 150 data as commented alternative for comparison - All varieties now consistently use threshold 172
Convert threshold from absolute value to percentage of actual pixel range. User inputs 0-255 (e.g., 172 = 67.45%), applied as: low + (high - low) * percentage. Provides consistent detection across varying lighting conditions.
Re-processed reference images using AI background removal (rembg) to fix pixel range detection issues caused by white/gray backgrounds.
Yaml starch scales
Moved duplicated QR/barcode metadata block into shared _add_qr_metadata() method in Analysis.py. Replaces 12 repeated lines in each of 4 analyses with a single method call.
All 4 varieties have monotonically decreasing ratings and were calibrated from MSU reference images using threshold 172.
Testing on 15 MSU starch reference varieties showed that 55% threshold achieves perfect monotonicity (15/15 varieties) compared to 67.45% which had 8/15 monotonic with 9 violations. The percentage-based threshold adapts to each image's actual pixel range, providing more consistent results across varying lighting conditions.
- Updated 12 MSU varieties with new 55% threshold data achieving 15/15 monotonicity with 0 violations - Added 5 new varieties: HONEYCRISP, BRAEBURN, EVERCRISP, GALA, ROME - All MSU varieties now use consistent integer index scales (1-8) - Kept non-MSU varieties unchanged (HONEY_CRISP, WA38, ALLAN_BROS, CORNELL, PURDUE, DANJOU, MINNEISKA, ROYAL_GALA)
Qr/Bar code detection
Add functional tests for QR detector, metadata parsing, and tray summary
Fix FileDirValue to validate input before setting value
Major corrections: - Analysis: Document _preRun(), _processImage(), _postRun() pattern instead of outdated performAnalysis() override pattern - Image class: Fix method names (getImageName not getFileName, addValue not addMetadata) - ImageIO: Clarify it's abstract with loadImage/saveImage for single images; ImageListValue handles lists via readValue/writeValue - GrannyUI: Clarify addProgramArgs() is not abstract (only run() is) - Scheduler: Fix schedule() return type (returns IDs, run() executes) - Update all code examples to use correct patterns
Fix developer guide docs
…EY_CRISP StarchArea.py: - extractImage() -> getPixelRange() - adjustImage() -> remapToRange() - gray -> grayscale - Removed unused grayscale_normalized variable starch_scales.yml: - Removed duplicate HONEY_CRISP entry (HONEYCRISP with 55% calibration is kept)
…tarch - Remove getPixelRange nested function and inline its histogram logic - Remove remapToRange function (dead code - was only used for visualization) - Remove unused grayscale_normalized variable - Clean up comments
…hold - CORNELL: Updated ratings from new calibration - PURDUE: Updated ratings from new calibration - DANJOU: Updated ratings, interpolated 1.5 and 3.5 for monotonicity, added 6.0 - WA38: Added new entry with averaged A/B sample ratings
Starch scales 55 percent
- WA38_1: New floral pattern scale (55% threshold), interpolated for monotonicity - WA38_2: New radial pattern scale (55% threshold), interpolated for monotonicity - ENZA: New scale (replaces incorrectly labeled ALLAN_BROS) - Removed: old WA38_1, WA38_2, WA38, ALLAN_BROS, MINNEISKA, ROYAL_GALA
Update starch scales: add ENZA, new WA38_1/WA38_2, remove old entries
Tests cover: YAML scale loading, StarchScales class attributes, parameter defaults and validation, _calculateStarch threshold math with synthetic images, _calculateIndex closest-match logic, and _drawMask pixel overlay behaviour. Grows test count from 54 to 76.
Add comprehensive unit tests for StarchArea
Raises the default YOLO confidence threshold from 0.25 to 0.7 to reduce false positive detections during segmentation. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Increase default segmentation confidence threshold to 0.7
The canonical file is Granny/assets/starch_scales.yml which has the current 55% threshold calibration data. The config/ copy was an old leftover from the initial YAML refactor.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This PR brings dev into main, incorporating 10 months of development work across six areas: CI/CD automation, starch scale overhaul, QR/barcode variety tracking, segmentation improvements, test expansion, and developer documentation fixes.
Changes
Starch Scale Overhaul
conditions
QR Code & Barcode Integration
Segmentation Improvements
CI/CD Pipeline
Test Expansion
Developer's Guide
Test plan