🤖 Model 🖼️
m1guelpf/nsfw-filter
Detect NSFW content in images. Takes an image and returns a moderation result with nsfw_detected (boolean), a list of NS...
← Back to search
NSFW Image Detection AI Models
NSFW image detection models classify images as safe or unsafe for public display. They are used for content moderation on platforms, user-generated content filtering, and automated safety checks in image processing pipelines.
How these models work
- Take an image as input and return a safety classification
- Most output a confidence score (0–1) so you can set your own threshold
- Some provide category-level breakdown (nudity, violence, drugs, etc.)
These models are essential for any platform that accepts user-uploaded images and needs to enforce content policies automatically.
Found 12 models (showing 1-12)
🤖 Model 🖼️
falcons-ai/nsfw_image_detection
Detect NSFW content in images, returning a binary label ('normal' or 'nsfw'). Accepts a single image as input and output...
🤖 Model 🖼️
jigsawstack/nsfw
Detects inappropriate content in images including nudity, explicit content, hentai, violence, and other NSFW categories....
🤖 Model 🖼️ → 📝
lucataco/flux-content-filter
Analyzes images and text prompts to detect content policy violations including public figures, copyright concerns, and N...
🤖 Model 🖼️
fofr/nsfw-model-comparison
Detect NSFW content in images and compare results across two classifiers. Takes an image input and returns JSON safety f...
🤖 Model 🖼️ → 📝
kojott/content-moderation-vision
Analyzes images for inappropriate content using MiniCPM-V-2.6 and returns structured safety assessments with detailed cl...
🤖 Model 🖼️
google-deepmind/shieldgemma-2-4b-it
Classify images for policy violations across sexually_explicit, dangerous_content, and violence_gore categories. Takes a...
🤖 Model 🖼️
meta/llama-guard-3-11b-vision
Classify the safety of multimodal inputs (image and user message) for content moderation. Accepts an image (required) an...
🤖 Model 🖼️
zsxkib/stable-diffusion-safety-checker
Detect NSFW content in images. Accepts an image and outputs a safety label indicating whether the content is safe or uns...
🤖 Model
meta/llama-guard-4-12b
Moderate text prompts, model responses, and images for safety compliance. Accepts text with optional multiple images and...
🤖 Model 🖼️
zsxkib/wd-image-tagger
Tag images with general, character, and rating labels from a single image input. Return a list of tags with confidence s...
🤖 Model 🖼️
diffground-oss/imaginet-detect
Detects whether an image is synthetic (AI-generated) or real, and identifies which AI model was likely used to generate...
No detection model is perfect. Set your confidence thresholds based on your platform's risk tolerance — stricter thresholds catch more violations but also flag more false positives. For production systems, consider combining model output with human review for borderline cases.