Built a tool for anyone drowning in huge image folders: HybridScorer
A free, local tool uses GPU and AI to score images by prompt match and aesthetic quality.
Developer vangel76 has released HybridScorer, an open-source tool designed to solve the tedious problem of manually sorting through massive folders of AI-generated or photographic images. The application runs entirely locally on a user's GPU, leveraging AI models to automatically score images based on two key criteria: how well they match a text prompt and their overall aesthetic quality. This automated scoring provides a powerful first-pass filter, drastically reducing the hours typically spent on manual curation.
After the AI scoring, HybridScorer presents the results in an interface that allows users to quickly review and manually correct any edge cases or mis-scored images. The final workflow lets users export clean, organized 'selected' and 'rejected' folders without altering the original files. A major technical benefit is its self-contained installation; it sets up all necessary dependencies within its own virtual environment, eliminating the common 'Python pain' and dependency conflicts that plague many open-source tools, making it accessible even for non-developers.
- Scores images locally using GPU AI for prompt match and aesthetic quality
- Creates clean selected/rejected folders without modifying original files
- Self-contained install in a virtual environment avoids Python dependency conflicts
Why It Matters
Saves hours for creators and researchers who need to curate large image sets, offering a free, local alternative to cloud services.