README.md

---
title: AI Image Checker MVP
emoji: 🖼️
colorFrom: blue
colorTo: green
sdk: streamlit
sdk_version: "1.32.0"
app_file: app.py
pinned: false
---

# AI Image Checker (MVP)

**What it does**
- Checks **provenance** via one-click link to **Adobe Verify** (upload the same image there to see Content Credentials / C2PA).
- Runs **EXIF** checks (missing/odd tags).
- Runs **ELA** (Error-Level Analysis) and a **lightweight ViT classifier**.
- Combines model + heuristics into a simple, explainable verdict.

**Deploy in 3 steps**
1) Create a **Hugging Face account** → New **Space** → Type: **Streamlit**.  
2) Upload these files: `app.py`, `requirements.txt`, `README.md`.  
3) Hit **Deploy**. That’s it.

**Use**
- Upload an image.
- Click **Open Adobe Verify** and upload the same image there to check for **Content Credentials** (the gold-standard when present).
- Review **EXIF**, **ELA**, and **AI confidence**. Adjust the threshold in the sidebar.

**Notes**
- Results are probabilistic; newer generators may fool detectors. Provenance (C2PA) > detection where available.
- This is a portfolio demo; not a certified forensic tool.