Counterfeit goods are becoming increasingly sophisticated, and traditional manual inspection is no longer enough to protect brands and customers. Today’s retailers need a fast, accurate, and scalable way to verify authenticity across branches, staff, and product lines.

At Simplico Co., Ltd., we developed an AI Authenticity Verification System designed to solve this challenge with cutting-edge machine learning, guided imaging, and automated reporting. This system empowers luxury brands, product resellers, and retail inspection teams to validate items with confidence using only a mobile device.

In this article, we share the complete overview of how the system works, the technology behind it, and why it represents the future of authenticity verification.

Why Brands Need AI Authentication

Manual inspection relies heavily on human expertise, leading to:

Inconsistent judgment between branches
Slow authentication time
Hard-to-train new staff
No visual proof or audit trail
Limited ability to scale

AI upgrades this process with speed, consistency, and explainability.

How the System Works

Our platform combines mobile apps, AI inference, backend processing, and a dashboard into a smooth workflow.

Diagram 1 — High-Level System Overview

flowchart TD

A["Staff Mobile App"] --> B["Secure Image Upload"]
B --> C["Django Backend API"]
C --> D["Job Queue (Celery)"]
D --> E["AI Inference Server (FastAPI + PyTorch)"]

E --> F["Heatmap Generator"]
E --> G["Confidence Score"]
F --> H["Object Storage (Images + Heatmaps)"]
G --> C

C --> I["PDF Report Generator"]
I --> H

C --> J["Dashboard & Analytics"]

1. Mobile App for Guided Image Capture

The mobile application ensures high-quality image collection from every angle. This improves data consistency and boosts AI accuracy.

Features:

Step-by-step guided capture
Auto-check for blur, lighting, alignment
AR overlay (Premium)
Auto-angle detection & auto-shutter
Secure uploads via pre-signed URLs

Diagram 2 — Mobile App Capture Flow

flowchart TD

A["Open Camera"] --> B["Guided Capture Steps"]
B --> C["Image Quality Scoring"]
C --> D{"Quality OK?"}
D -->|No| B
D -->|Yes| E["Secure Upload to S3/MinIO"]
E --> F["Submit Verification Job"]

2. AI Authenticity Engine

The core intelligence of the platform analyzes product patterns such as stitching, logo geometry, textures, and hardware structure.

Three AI Levels:

Basic Model — For pilot/initial rollout
Pro Model — Better feature extraction
Ensemble + YOLOv8 — Highest accuracy, multi-layer explainability

Outputs include:

Classification (authentic / suspicious / uncertain)
Confidence score
Heatmap visualization

Diagram 3 — AI Processing Pipeline

flowchart TD

A["Received Images"] --> B["Preprocessing"]
B --> C["AI Model (Basic / Pro / Ensemble)"]
C --> D["Feature Extraction"]
D --> E["Prediction + Confidence Score"]

E --> F["Grad-CAM / Heatmap Generation"]
F --> G["Upload Heatmaps to Storage"]

E --> H["Return AI Results to Backend"]

3. Backend (Django, Celery, PostgreSQL)

The backend orchestrates all processes:

Authentication & role permissions
Job queue & async processing
PostgreSQL structured storage
PDF report generation
Staff, branch & SKU analytics
Audit logging

Diagram 4 — Backend and Inference Interaction

sequenceDiagram
    autonumber
    participant App as "Mobile App"
    participant API as "Django Backend"
    participant Q as "Celery Queue"
    participant AI as "AI Server"
    participant DB as "PostgreSQL"
    participant S3 as "Object Storage"
    participant PDF as "PDF Generator"

    App->>API: Submit verification request
    API->>S3: Store raw images
    API->>Q: Create inference job
    Q->>AI: Run AI model
    AI->>S3: Upload heatmaps
    AI->>Q: AI results
    Q->>API: Save results
    API->>DB: Insert record
    API->>PDF: Generate PDF report
    PDF->>S3: Upload PDF
    API-->>App: Return final result

4. PDF Report

Each authentication process generates a professionally formatted PDF containing:

All captured images
Heatmap overlays
Confidence score
AI explanation
Timestamp, branch, staff
Secure share link

This builds trust with customers and ensures compliance.

5. Dashboard & Analytics

The dashboard transforms raw operations into business insights.

Branch Analytics

Volume of verifications
Pass/fail/uncertain distribution
Staff activity patterns

Staff Analytics

Capture consistency
Verification speed
Training recommendations

Advanced (Premium)

SKU-level trends
Model drift monitoring
Risk scoring

Diagram 5 — Analytics Architecture

flowchart TD

A["Backend Logs"] --> B["Analytics Processor"]
B --> C["Branch Analytics Dashboard"]
B --> D["Staff Performance Dashboard"]
B --> E["SKU Risk Dashboard"]

Deployment Options

Cloud (Recommended)

Scalable GPU
Lower maintenance
Faster rollout
15,000–20,000 THB/month

On-Premise

Local inference
Higher upfront cost
Requires cooling & maintenance

Package Overview

Category	Small	Medium	Premium
App	Android	Android	Android + iOS
Capture Workflow	Manual	Guided	AR + Auto
AI Model	Basic	Pro	Ensemble + YOLO
Explainability	Single	Multi	Multi-layer
PDF Reports	No	Yes	Enhanced
Analytics	Basic	Full	Advanced