Developer Infrastructure

Visual Content Moderation & Precise Censorship

Shield your application from harmful visual content. Safely scan uploads for nudity, gore, weapons, and blood, then automatically blur or block offending regions with a single click.

Try the Moderation Playground

Upload an image or click a sample preset. Toggle censorship parameters on the fly.

1. Upload & Configure

Select one of our simulated triggers to test coordinates and outputs:

Safe Scene

Gore (Red)

Weapon

Blood

NSFW

Drag & Drop or click to upload your custom image

Max file size: 5MB. Supports JPG, PNG, WEBP.

Select Censorship Filter Mode:

Censor entire image instead of regions

2. Interactive Output

Response JSON

{
  "info": "Upload or click Scan to analyze this image."
}

Designed for High Scale Developers

Complete visual governance stack built for sub-100ms API responses.

Separated Workflows

Scan images to retrieve coordinates, then choose to apply blur, pixelate, or black boxes on the fly or host custom reviews.

Multi-Class Bounding Boxes

Detect exact locations of weapons, nudity, and bleeding wounds to blur only the violating portions while keeping rest visible.

Sub-50ms Latency

Triton inference acceleration allows real-time scanning pipelines optimal for chat programs and online community boards.

Scan & Filter Profiles

Comprehensive category analysis for robust content filtering.

NSFW & Nudity

Granular classification for adult content, anatomy exposures, and highly suggestive material to protect user-facing grids.

Gore & Disturbing

Identifies medical trauma, open incisions, accidents, and horrifying scenes that induce community discomfort.

Weapons & Firearms

High recall object detection models for pistols, rifles, blade edges, and tactical gear to filter threatening contexts.

Blood & Injury

Detects large-scale bleeding, lacerations, and violence-related crimson distribution mapping coordinates.

Stream Ingestion (Future)

Designed with live RTMP frame capture pipelines for real-time video stream inspection overlays.

SDK-First Design

Equipped with standardized open wrappers for Python, Node.js, and raw cURL inputs.

Developer Documentation

Quick start guides, API endpoints references, and SDK integrations.

Getting Started

SafeVision is a content moderation API designed for security, separation of concerns, and developer ease. To make API requests, you need to sign up for a console account and generate an API key.

Authorization Header

All requests to the scan or censor endpoints must authenticate using the X-API-Key header:

X-API-Key: sv_live_your_secret_key_here

Scan API Reference

Analyze an image to evaluate safety confidence, tag classifications, and bounding boxes.

POST /api/v1/moderation/scan

Request Body (Multipart Form)

Param	Type	Required	Description
`file`	File (Binary)	Yes	The JPEG, PNG, or WEBP image to examine. (Max size: 20MB)

Query Parameters

Param	Type	Required	Description
`model` Pro	String	No	Override the Hawk detector version for this specific request. Accepted values: `v1`, `v2`, `latest`. Overrides your saved dashboard preference. Ignored on Sandbox plan.

Sample Output JSON

{
  "safe": false,
  "categories": [
    {
      "type": "weapon",
      "confidence": 0.94,
      "severity": "high",
      "box": {
        "x": 120,
        "y": 80,
        "width": 250,
        "height": 180
      }
    }
  ],
  "image_hash": "e3b0c442...",
  "latency_ms": 32,
  "model": {
    "detector": "Hawk v2",
    "classifier": "v1"
  }
}

Censor API Reference

Apply local censorship filters (blur, pixelate, solid box) to the identified bounding regions of an image.

POST /api/v1/censor/apply

Request Body (Multipart Form)

Param	Type	Required	Description
`file`	File (Binary)	Yes	The image to process.
`moderation_results`	String (JSON)	No	Scan categories containing bounding boxes. If omitted, the server will scan it automatically.
`mode`	String	No	`blur`, `pixelate`, or `black_box`. (Default is `blur`)

Developer SDKs

We provide official lightweight wrappers to implement visual moderation with minimal overhead.

Python Integration

# Install the SDK
# pip install safevision

from safevision import SafeVision

# Initialize client
sv = SafeVision(api_key="sv_live_xxxx")

# Scan with default model (uses your dashboard setting)
result = sv.scan("photo.jpg")
print(result["safe"])        # True or False
print(result["categories"])  # bounding boxes
print(result["model"])       # {"detector": "Hawk v2", "classifier": "v1"}

# Pro: Override Hawk model version per-request
result_v2 = sv.scan("photo.jpg", model="v2")  # Force Hawk v2
result_v1 = sv.scan("photo.jpg", model="v1")  # Force Hawk v1 (legacy)

# Apply censorship filter
censored_bytes = sv.censor("photo.jpg", result, mode="blur")
with open("censored.jpg", "wb") as f:
    f.write(censored_bytes)

JavaScript Integration

import { SafeVision } from 'safevision-node';

const sv = new SafeVision({ apiKey: 'sv_live_xxxx' });

// Scan with default model
const result = await sv.scan('./upload.png');
console.log(result.safe ? 'Clean!' : 'Harmful content detected');
console.log(result.model); // { detector: 'Hawk v2', classifier: 'v1' }

// Pro: Override Hawk model version per-request
const resultV2 = await sv.scan('./upload.png', { model: 'v2' }); // Force Hawk v2

// Get blurred image
const censoredBuffer = await sv.censor('./upload.png', { mode: 'pixelate' });

Raw HTTP Example (curl):

curl -X POST "https://safevision.guardextech.com/api/v1/moderation/scan?model=v2" \
  -H "X-API-Key: sv_live_xxxx" \
  -F "file=@photo.jpg"

Flexible, Developer-Friendly Pricing

Choose the billing plan that aligns with your request volumes.

Developer Sandbox

Ideal for prototyping & testing integration pipelines.

1,000 free scans / month
Standard blur & pixelate modes
SQLite default logs tracking
Single active API key prefix

5-Day Free Trial

Professional Plan

Includes 5-day free trial on email verification!

$29 / mo

50,000 scans / month
Ultra-low latency inference
Custom bounding box overrides
Multi-category detection tuning
5 concurrent API keys

Enterprise Scale

For large-scale community hubs requiring dedicated GPUs.

Custom

Uncapped scans / month
Dedicated Triton GPU clusters
RTMP live-stream filtering (1s feed)
On-premise deploy (Docker/K8s)
99.99% SLA uptime agreements

Get in Touch

Have a question, a custom request, or need help? We're here for you.

Email Support

Send us a message and we'll get back to you within 24 hours.

safevisioncontactsupport@gmail.com

Enterprise Inquiry

Need custom volume, dedicated GPU clusters or on-premise deployment? Let's talk.

Custom pricing · SLA agreements · Dedicated support

Developer Console

Manage API credentials, view scan analytics, and audit request history.

Total API Scans

Flag Rate

Avg Latency

0 ms

Active Plan

Sandbox

API Traffic (Last 7 Days)

Recent scan logs

Hash	Result	Latency	Matched Filters	Timestamp
No records found. Generate a key and scan an image to view logs.

Credentials Manager

Use credentials to authorize requests from code or terminal clients.

Active Keys

No active credentials. Generate one above.

Model Configuration

Select which version of the Hawk detector runs on your scans.

Latest (Automatic) Automatically routes to the newest stable Hawk detector model.

Hawk v2 (Retraining) Newest model with enhanced negative kitchen datasets.

Hawk v1 (Legacy) First generation detector trained on custom threats.

Account Settings

Update your developer account password.

Danger Zone

Manage account deletion and active subscription cancellation.

Delete Developer Account

Permanently delete your account, generated API keys, and logs.

Legal Documentation

Review our API Terms of Service and user Privacy Policy guidelines.

Terms of Service

Last updated: June 24, 2026

Welcome to SafeVision. By accessing or using our Content Moderation API endpoints, developer console, or SDK integrations, you agree to comply with and be bound by the following Terms of Service. Please read them carefully.

1. License and Scope of Use

SafeVision grants you a limited, non-exclusive, non-transferable, revocable license to access our APIs to scan, filter, and censor image content for your applications. Sandbox plans are limited to 1,000 monthly scans, and any automated stress-testing or abuse of endpoints will result in account suspension.

2. Content Restrictions

Our detector models are designed to identify weapons, blood, gore, and adult suggestive content (NSFW) to safeguard communities. However, you may not use our API to store, index, or parse illegal content, child exploitation material, or distribute copyrighted images without consent.

3. Liability and Warranties

SafeVision is provided "as is" and "as available". We do not guarantee 100% classification accuracy or that the detector will catch all threats. SafeVision / Guardex Tech Labs is not liable for any content moderation failures or resulting user experience disruptions in your software.

Visual Content Moderation & Precise Censorship

Try the Moderation Playground

1. Upload & Configure

2. Interactive Output

Designed for High Scale Developers

Separated Workflows

Multi-Class Bounding Boxes

Sub-50ms Latency

Scan & Filter Profiles

NSFW & Nudity

Gore & Disturbing

Weapons & Firearms

Blood & Injury

Stream Ingestion (Future)

SDK-First Design

Developer Documentation

Getting Started

Authorization Header

Scan API Reference

Request Body (Multipart Form)

Query Parameters

Sample Output JSON

Censor API Reference

Request Body (Multipart Form)

Developer SDKs

Flexible, Developer-Friendly Pricing

Developer Sandbox

Professional Plan

Enterprise Scale

Get in Touch

Email Support

Enterprise Inquiry

Verify Your Email

Developer Console

API Traffic (Last 7 Days)

Recent scan logs

Credentials Manager

Copy Key Securely

Active Keys

Model Configuration

Pro Feature

Account Settings

Danger Zone

Legal Documentation

Terms of Service

1. License and Scope of Use

2. Content Restrictions

3. Liability and Warranties

Privacy Policy

1. Information We Collect

2. Moderation Request Images

3. Account Deletion Rights (GDPR / KVKK)

SafeVision Billing

Professional Plan

Payment Verified!

Confirm Downgrade

Downgrade to Developer Sandbox?

Delete Developer Account