ViT NSFW Stable Diffusion Detector
Upload an image — get an NSFW / SAFE verdict with confidence scores, powered by AdamCodd/vit-nsfw-stable-diffusion, a ViT fine-tuned on Stable Diffusion outputs (93.5% accuracy).
Deployment status
1
Model access not yet approved. The model repo is manually gated.
Email
adamcoddml@gmail.com with a short project description to request access.2
Compute hosting not enabled. Gradio Spaces need ZeroGPU (free for accounts older
than 30 days) or a PRO plan. Once eligible, set
zero-a10g hardware and add the approved
account’s token as the HF_TOKEN secret.Go live in three steps
1
Get approval for
AdamCodd/vit-nsfw-stable-diffusion (email the author).2
Enable hosting: ZeroGPU (
zero-a10g) on a 30-day-old+ account, or subscribe to PRO.3
Push the Gradio app (
app.py, README.md with sdk: gradio,
requirements.txt, examples/) and set the HF_TOKEN secret.About the app
- Gradio UI with upload / webcam / clipboard input.
- Colored NSFW / SAFE banner, top-3 confidence bars, and raw JSON scores.
- Runs on ZeroGPU without burning GPU quota (classification is CPU-bound).
- Exposes an MCP server (
mcp_server=True) for API use.
curl https://huggingface.co/new-space/agents.md