← All projects

Trust & Safety · Computer Vision

AI-Generated Image and Deepfake Detection

A planned detection platform for AI-generated imagery and face manipulation.

Product exploration

Overview

A product-oriented research initiative exploring detection of AI-generated images and face manipulation using pretrained vision encoders, task-specific training, and a measurable evaluation pipeline.

Methods & topics

Visual forensicsPretrained vision encodersBinary classificationRobustness evaluationAPI design

Highlights

  • Defining separate detection tasks for generated images and face manipulation.
  • Planning dataset curation, provenance tracking, and evaluation protocols.
  • Exploring a web experience and API for model inference.