smart_asset_analyser 0.1.6

A Flutter package to detect visually identical and similar assets using Deep Visual Embeddings (CLIP). Supports images, SVGs, and Lottie animations with interactive HTML reports.

Changelog

All notable changes to this project will be documented in this file.

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

0.1.6 - 2025-01-27

Fixed

• Fixed dependency detection to check for transformers instead of clip-by-openai
• Updated checkDependencies() to correctly detect installed Python packages
• Added fallback check for backwards compatibility

0.1.5 - 2025-01-27

Fixed

• CRITICAL: Removed clip-by-openai dependency which had conflicts with PyTorch 2.0+
• Switched to transformers library (Hugging Face) for CLIP implementation
• Updated all Python code to use transformers.CLIPModel instead of clip-by-openai
• Fixed dependency resolution errors when installing from pub.dev

Changed

• Updated requirements.txt to remove clip-by-openai
• Updated all documentation to reflect new dependency requirements
• Python code now uses Hugging Face transformers for better compatibility

0.1.4 - 2025-01-27

Added

• Requirements helper with automatic requirements.txt detection
• --show-requirements flag to display installation instructions
• Automatic pip vs pip3 detection

0.1.3 - 2025-01-27

Fixed

• Fixed Python script location when installed from pub.dev
• Updated package name references

0.1.2 - 2025-01-27

Fixed

• Fixed executable name in pubspec.yaml
• Renamed binary file to match executable name

0.1.1 - 2025-01-27

Changed

• Package name updated to smart_asset_analyser
• Updated all documentation to use version dependencies (^0.1.0) instead of path dependencies
• Improved Python script location resolution using package_config for pub.dev compatibility
• Updated CLI command to dart run smart_asset_analyser:analyse assets

Added

• Example Flutter project in example/ folder
• Comprehensive example project documentation
• Better error messages for Python script location

Fixed

• Python scripts now work correctly when package is installed from pub.dev
• All package references updated to smart_asset_analyser
• CLI argument parser fixed

0.1.0 - 2025-01-27

Added

• Initial release of Smart Asset Analyser
• Deep Visual Embeddings using CLIP model via Python bridge
• Support for multiple asset types:
  • Images (PNG, JPG, JPEG, WebP, GIF, BMP)
  • SVGs (rasterized for comparison)
  • Lottie animations (key frames extracted)
• Interactive HTML report with filtering:
  • Similarity percentage slider (0-100%)
  • Asset type filter (Images, SVGs, Lottie)
  • Search by filename
• CLI command: dart run smart_asset_analyser:analyse assets
• Configurable similarity thresholds
• Embedding caching for faster subsequent runs
• Batch processing support
• HTTP server mode for faster processing
• Comprehensive documentation

Technical Details

• Uses CLIP ViT-B/32 model for visual embeddings
• Cosine similarity calculation for comparison
• Union-find algorithm for grouping similar assets
• Python bridge for CLIP model integration
• SVG rasterization using cairosvg
• Lottie frame extraction using lottie