vangel76

vangel76 / HybridScorer

Public

Quick AI assisted human-in-the-loop image triage.

clip image imagereward scoring
49
3
100% credibility
Found Apr 10, 2026 at 49 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

HybridScorer is a local desktop app that uses AI to score and sort large image folders into selected and rejected groups based on text descriptions for content or aesthetics, with simple visual review and export options.

How It Works

1
🔍 Overwhelmed by image piles

You have tons of pictures from AI creations or collections, but sorting them manually feels endless.

2
📥 Grab the helper tool

Download the project folder and run the simple setup clicker for your Windows or Linux computer.

3
🚀 Launch your sorter

Click the run helper, and a friendly webpage opens right on your screen, ready to go.

4
Choose your sorting style
🎯
Match a description

Type words like 'blonde woman in dress' to find pictures that fit your vision.

⭐
Pick the best looks

Describe the style or mood you love to rank pictures by overall appeal.

5
⚡ Let it score everything

Enter your words, press go, and it zips through your folder splitting keepers from rejects.

6
👀 Check and fix easily

Browse the two groups of picture previews, drag mistakes to the right side, and slide the cutoff.

7
📂 Export your favorites

Pick which groups to save, name new folders, and copy the winners without touching originals.

✅ Perfect sorted folders

Celebrate your clean, clutter-free collections of just the images you want to keep.

Sign up to see the full architecture

6 more

Sign Up Free

Star Growth

See how this repo grew from 49 to 49 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is HybridScorer?

HybridScorer is a Python tool for quick, assisted human-in-the-loop image triage on large folders. Point it at your images, score them via GPU-accelerated AI for content matching (like finding dupes or prompt fits) or aesthetic appeal, then manually tweak the selected/rejected split in a Gradio web UI before exporting clean buckets. It solves the slog of hand-sorting AI generations, references, or photo dumps without touching originals.

Why is it gaining traction?

Setup is dead simple with one-line scripts for Windows/Linux virtualenvs and CUDA PyTorch—quick github setup if you've done this kind of thing before. Users love the seamless switch between PromptMatch for semantic search and ImageReward for style ranking, plus proxy caching for fast rescoring and histogram previews for threshold tweaks. Local, private runs with no cloud mean quick github share of results, beating hosted tools on control and speed.

Who should use this?

AI artists culling thousands of Stable Diffusion outputs for keepers. Designers quick-looking reference boards to match subjects, poses, or moods. Photographers or collectors doing github quick dupe removal or github quick search on local archives via custom prompts.

Verdict

Grab it for niche image workflows—49 stars and 1.0% credibility score signal early maturity, but polished docs and quick github tutorial make it low-risk to spin up. Solid if you need fast triage now; watch for broader testing as it grows.

(187 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.