Deepfake Image Detection Tool
A tool that judges whether an image involves face swapping or manipulation. Usable as a web app, and also callable from other systems via a REST API.
Duration
Short-term development
Team
1 person
01 // OVERVIEW
Project Overview
A system that judges whether an image is a deepfake (a face swap or manipulation). Users can upload an image from a browser for judgment, and it can also be embedded in other systems as a REST API. It integrates with an external judgment API and returns results in three levels - safe, warning, or danger - based on the score.
02 // KEY FEATURES
Key Features
Image Upload Judgment
Accepts JPEG, PNG, and WebP images and scores the likelihood of a deepfake. Displays results in three levels: safe, warning, or danger.
REST API
Provides an endpoint that accepts images via multipart/form-data, supporting calls from other systems. Rate limiting is also implemented.
Docker Operation
Can be launched with Docker Compose, supporting switching between development and production environments.
03 // TECHNICAL DEEP-DIVE
Technical Highlights
Solution
Safely Integrating an External Judgment API
Managed the API key via environment variables and added file size and format validation plus rate limiting, creating a setup robust enough for public use.
04 // IMPACT & RESULTS
Results
3 Levels
Safe, warning, danger
API-Ready
Embeddable in other systems