v0.2.0

What's Changed

• New measure command: Batch-compute morphology metrics (area, perimeter, body length, body width, Feret diameters, etc.) from segmentation masks. Exports metrics.csv, metrics_summary.csv, and metric_definitions.csv. Includes a caution notice that body length/width are mask-geometry estimates and may be inaccurate when masks include appendages, touch borders, or contain merged/fragmented regions.
• ONNX progress bar: Added progress bar for ONNX inference in classify predict and classify evaluate.
• Segment annotation alignment: Annotation granularity now matches --segmentation-method: bbox-only methods output bounding-box annotations, while full segmentation methods output both bounding-box and segmentation polygons with mask-area.

Documentation

• Updated English and Chinese README with measure usage and caution notes.
• Updated entomokit-workflow skill (SKILL.md, workflow.md, command-profiles.md, teaching-playbook.md) to cover the new measure command and its interpretation guardrails.

v0.1.7

v0.1.7 — Bug Fix Release

Bug Fixes

• split-csv class count CSVs: Fixed a bug where all class.*.count files (train, val, test.known, test.unknown) were being written as single-column CSV files (only count, missing label column). Root cause: value_counts().to_csv(index=False) silently dropped the label column from the Series index. Fixed by adding a _write_class_count() helper that consistently produces label,count columns.

v0.1.6

What's Changed

Segmentation Improvements

• Fixed Otsu segmentation: Now evaluates both BINARY and BINARY_INV threshold directions, picks the better candidate to avoid full-image foreground collapse on light backgrounds
• Fixed Otsu color conversion: Corrected RGB→GRAY conversion (was incorrectly using BGR2GRAY on RGB images from PIL)
• Fixed GrabCut scaling: Rescales mask to original image dimensions when internal downscaling was applied for large images

New Features

• Added otsu-bbox and grabcut-bbox segmentation methods for fast RGB crop output (no alpha channel)

Documentation

• Updated README and skill docs with all current segmentation methods: sam3, sam3-bbox, otsu, otsu-bbox, grabcut, grabcut-bbox

v0.1.5

New Features

AI Assistant Integration (entomokit-workflow skill)

Added a skill for AI assistants (OpenCode, Claude Code, Codex) to guide non-CLI users through the EntomoKit pipeline with conversational workflow orchestration.

Key capabilities:

• Guided workflow through all phases (doctor → clean → split-csv → classify)
• Parameter validation against CLI schema before execution
• CSV teaching and validation
• Error recovery with actionable suggestions
• Demo mode with repository sample data

v0.1.4

New Features

entomokit doctor

Environment diagnostics command to check and report system configuration issues.

entomokit augment

Data augmentation pipeline for image datasets.

Improvements

• Updated README with better documentation and examples
• Added dual deployment guidance (PyPI + skill)
• Solidified skill MVP scope and acceptance gates
• Added skillization design documentation for non-CLI users

v0.1.3

Initial release of entomokit - A Python toolkit for building insect image datasets