Age-X: Next-Generation Age Verification Platform

Executive Summary

Age-X is a safety-critical, high-performance web application designed to protect minors from age-inappropriate content in real-time. Unlike traditional age gates, Age-X utilizes on-device camera input and server-side neural inference (ResNet18) to continuously monitor the viewer's age group and filter content dynamically.

This repository contains the production-ready source code for the platform, engineered with a "Safety-First" architecture where fallback states always default to the most restrictive content category (Kid Mode).

---

🏗️ Architecture

High-Level Design

The system follows a decoupled Client-Server architecture:

1. Frontend (Vanilla JS + GSAP): A lightweight, accessible client that manages the camera feed, handles secure video playback, and renders a glassmorphism-based UI.
2. Backend (FastAPI): A stateless, high-throughput REST API that provides inference services using a loaded PyTorch model.
3. Safety Engine: Both client-side and server-side logic enforce "Strict Unlocking", where upgrading to Adult privileges requires sustained high-confidence verification, while downgrading to Kid mode is instantaneous.

Directory Structure

Age-X/
├── backend/
│   ├── app.py              # Application Entry Point (FastAPI)
│   ├── age_service.py      # Core Inference Logic & Model Singleton
│   ├── age_model.pth       # Neural Network Weights (ResNet18)
│   └── requirements.txt    # Python Dependencies
├── frontend/
│   ├── index.html          # Semantic Entry Point
│   ├── style.css           # Premium CSS / Glassmorphism System
│   ├── app.js              # Application Logic (Safety Engine, Reel Manager)
│   └── firebase.js         # Content Database Config
└── README.md               # Documentation

---

🔒 Security & Privacy Protocol

Data Handling

• Ephemeral Processing: Images sent to /api/age-check are processed in memory and typically discarded after inference. We do not store face data on disk.
• No PII Storage: No facial embeddings or personally identifiable information is persisted.
• HTTPS Enforcement: In production, all API calls must be over TLS 1.3.

Safety Logic (Strict Mode)

The SafetyEngine implements a state machine with the following rules:

• Default State: Kid (Locked).
• Upgrade Path: Kid -> Adult requires 5 consecutive frames of >85% confidence.
• Downgrade Path: Adult -> Kid occurs on 1 single frame of detection (Zero Tolerance).
• Fail-Safe: Network errors, camera failures, or ambiguous results force Kid mode.

---

🚀 Setup & Deployment

Prerequisities

• Python 3.10+
• Node.js (optional, for local serving)
• Webcam

Installation

1. Backend Setup:

   cd backend
   pip install -r requirements.txt

2. Start API:

   uvicorn app:app --reload --host 0.0.0.0 --port 8000

3. Frontend Setup:
   • Serve the frontend directory using any static file server (e.g., Live Server in VS Code):

   cd frontend
   npx serve .

Model configuration

Ensure age_model.pth is present in backend/. If missing, the system will initialize but fail open to Safety Mode (Kid) for all requests. The model should be a standard ResNet18 trained for age regression.

---

🧪 Performance Targets

• Inference Latency: ~80-150ms on CPU (optimized with opencv-python-headless).
• UI Interaction: 60fps animations via GSAP.
• Responsiveness: Mobile-first design supporting viewports from 320px to 4K.

---

⚖️ License & Compliance

Proprietary software tailored for silicon-valley standard safety compliance.