Explore the world of AI-powered Deepfakes technology in this comprehensive 51-minute conference talk from Devoxx. Dive into the fundamentals of Deepfakes, including their creation tools and potential future developments. Gain insights into the positive applications and ethical considerations surrounding this technology. Learn about the underlying AI concepts and discover why specific AI models are employed. Examine the technical stack behind Faceswap, including Python, Keras, Tensorflow, and PlaidML, and understand the challenges of maintaining a public project using cutting-edge research technologies. Follow the Faceswap process from data gathering to final compositing, covering face detection, alignment, recognition, and segmentation. Conclude with a discussion on the future of Deepfakes and participate in a Q&A session to further expand your knowledge on this fascinating subject.
Faceswap - Developing a Platform for Practical Deepfakes
Overview
Syllabus
Intro
Overview
How did we get here?
Features of Faceswap
API Stack
Neural Networks
The Faceswap Machine Learning Stack
Face Detection
Face Alignment
Face Recognition
Face Segmentation
Image Reconstruction with Auto-Encoders
Final Steps: Compositing & Splicing
Getting Started with Faceswap
Faceswap Process
Gathering the Data
Extract
Convert
Post-processing previews
The future of Deeplakes
References
Conclusion and Q&A
Taught by
Devoxx
