[Interactive] Neural Style Transfer: The Art and Science of Generative AI
summary
Walks through neural style transfer that uses pre-trained CNN feature extractors like VGG or ResNet, adds explicit style and content loss functions, and iterates to produce one image that keeps edges/lines while transferring brush strokes and colours. Shows a live interactive demo where attendees submit content and style images (the demo runs and generates outputs), cites Prisma as a deployed example, lists challenges like compute and high‑resolution synthesis and names fixes such as multi-scale synthesis and style‑scale tuning for medical imaging and film/design use cases.
AI-written summary · about
chapters (6)
- 0:00Introduction to Generative AI
- 1:15Neural style transfer basics
- 2:27How convolutional networks work
- 6:48The underlying architecture
- 8:45Recap and technical process
- 10:18Applications and future use
related talks
58mStyle Transfer AI - How Neural Networks Generate Art
37mLearning AI with Van Gogh, Botero and Cruz‑Diez
25mMixing art with Python- an introduction to Style Transfer
40mLearn to be a painter using Neural Style Painting
50mVan Gogh Meets Neural Networks
28mApplications of Generative Modeling
06mGAN-based Style Transformation to Improve Gesture-recognition Accuracy
40mOverview of Generative Adversarial Networks (GANs)