v0.1.4

Fixed

• Fixed id conflicts when merging datasets with overlapping image IDs
• Fixed minor bugs

Changed

• Enhanced test CI workflow with coverage reporting

Added

• Added unit tests for all core functionalities

Full Changelog: v0.1.3...v0.1.4

v0.1.3

Fixed

• Fixed split error when exporting datasets in GUI annotator
• Fixed import errors

Added

• Added streamlit-based audit report analysis tool (experimental), accessible via annotator menu

Full Changelog: v0.1.2...v0.1.3

v0.1.2

Fixed

• Fixed image tags not being saved in GUI annotator
• Fixed image tags not displaying in CLI dataset info
• Fixed image panel focus issues when approving/rejecting images in GUI annotator

Added

• Added split ratio options for dataset export in GUI annotator

v0.1.1

Fixed

• Fixed broken documentation links throughout the documentation site
• Fixed Windows Poetry installation issues in GitHub Actions workflows
• Fixed ANSI color formatting in CLI output display
• Fixed mkdocs strict mode build failures

Changed

• Simplified CI workflows with direct pip installation of Poetry
• Improved CLI help formatter with better spacing and readability
• Updated documentation navigation structure
• Enhanced README with more comprehensive examples

Added

• Added changelog and license pages to documentation
• Added About section to documentation site
• Added better error handling for empty test suites in CI

v0.1.0

Added

Core Features

• Initial release of BoxLab
• Dataset management system with multi-source support
• COCO and YOLO format support
• Bidirectional format conversion (COCO ↔ YOLO)
• Dataset merging with conflict resolution
• Dataset splitting with customizable ratios
• Comprehensive statistics and analysis

CLI Tools

• boxlab dataset info - Display dataset information and statistics
• boxlab dataset convert - Convert between formats with flexible options
• boxlab dataset merge - Merge multiple datasets with source tracking
• boxlab dataset visualize - Generate visualizations and plots
• boxlab annotator - Launch GUI annotation application
• Rich formatted output with tables and progress indicators
• Verbose mode for detailed debugging

GUI Annotator

• Interactive annotation viewer and editor
• Visual bounding box editing with drag-to-resize
• Multi-split navigation (train/val/test)
• Audit workflow with approve/reject functionality
• Image tagging system
• Workspace persistence (.cyw files)
• Auto-backup on crashes
• Keyboard shortcuts for efficient workflows
• Real-time statistics display

Python API

• Dataset class for dataset management
• load_dataset() and export_dataset() functions
• Type-safe data structures (BBox, Annotation, ImageInfo)
• Plugin system for custom format support
• LoaderPlugin and ExporterPlugin base classes
• Plugin registry for format discovery
• Multiple naming strategies (original, UUID, sequential, prefix)

PyTorch Integration

• TorchDatasetAdapter for PyTorch Dataset compatibility
• build_torchdataset() convenience function
• Built-in augmentation support
• Flexible transform pipelines
• Custom collate functions
• Multiple coordinate formats (xyxy, xywh, cxcywh)
• Normalization options

Documentation

• Complete user guides (installation, quick start, CLI reference)
• Comprehensive API reference with examples
• Advanced usage guide (PyTorch, plugins, automation)
• Interactive annotator guide
• MkDocs Material theme documentation site

Development

• Poetry for dependency management
• Ruff for linting and formatting
• MyPy for type checking
• Pytest for testing
• GitHub Actions workflows (tests, docs, publish)
• Pre-commit hooks

Format Support

COCO Format

• Full COCO JSON specification support
• Image metadata preservation
• Category management
• Annotation area calculation
• Source tracking in merged datasets

YOLO Format

• YOLOv5/YOLOv8 format support
• data.yaml configuration
• Normalized coordinate handling
• Split-based directory structure
• Automatic path resolution

Plugin System

• COCOLoader and COCOExporter
• YOLOLoader and YOLOExporter
• Extensible architecture for custom formats
• Format auto-detection
• Validation hooks

Dependencies

• Python 3.10+ support
• NumPy for numerical operations
• Pillow for image handling
• Matplotlib for visualizations
• PyYAML for YAML parsing
• Pandas for data analysis
• Optional PyTorch integration

System Support

• Cross-platform (Linux, macOS, Windows)
• Python 3.10, 3.11, 3.12 support
• Terminal color support with auto-detection
• Graceful degradation without optional dependencies