7th Workshop on Computer Vision for Fashion, Art, and Design

Event: CVFAD @ CVPR 2024 · Duration: 183 min · ▶ Watch on YouTube

Abstract

The 7th Workshop on Computer Vision for Fashion, Art, and Design (CVFAD) at CVPR 2024 explores the intersection of computer vision and creative domains. This year’s workshop features keynote speakers and accepted papers covering topics from fine-grained fashion retrieval to 4D dynamic content creation. The event aims to bring together artists, designers, researchers, and engineers to discuss new perspectives and advancements in AI and machine learning for creative applications. The workshop also highlights the challenges of bias in AI models and the potential for new forms of artistic expression through computational tools.

Speakers

  • Karsten Kreis — NVIDIA
  • Nouf Aljowaysir — Independent New Media Artist
  • Jason Salavon — University of Chicago
  • Jia-Bin Huang — University of Maryland College Park
  • Ziad Al-Halal — University of Utah
  • Loris Bazzani — Amazon
  • Nour Karessli — Zalando SE
  • Julia Lasserre — Zalando SE
  • Leonidas Lefakis — Zalando SE
  • Negar Rostamzadeh — Google Research
  • Reza Shirvany — Zalando SE
  • Mariya Vasileva — AWS

Talks (5)

  • 00:00:00 — Loris Bazzani: Opening Remarks
    • Loris Bazzani introduces the 7th Workshop on Computer Vision for Fashion, Art, and Design, highlighting its focus on creative domains and the goal of bringing together artists, designers, researchers, and engineers.
  • 00:00:00 — Nouf Aljowaysir: The Simplification of Culture: An Artist Perspective on Computer Vision
    • Nouf Aljowaysir discusses her journey exploring AI in creative commercial spaces and from an intimate/personal perspective, focusing on how AI interprets cultural artifacts and the biases inherent in current models.
  • 00:00:00 — Karsten Kreis: Generative AI as Artists’ Paintbrush: From Diffusion Models to 4D Dynamic Content Creation
    • Karsten Kreis explores the capabilities of generative AI, particularly diffusion models, in creating diverse and complex visual content, including images, videos, and dynamic 3D objects, and discusses the technical advancements and challenges in this field.
  • 00:00:00 — Jia-Bin Huang: Exploring New Perspectives with Computer Vision
    • Jia-Bin Huang presents innovative computer vision techniques for creating immersive and interactive visual experiences, including 3D reconstruction from 2D images, dynamic wire art, and anamorphic art, emphasizing the potential for new forms of artistic expression.
  • 00:00:00 — Jason Salavon: Good Art is Off-Manifold
    • Jason Salavon discusses his artistic practice using generative AI, focusing on the creation of abstract paintings, image averages, and large-scale digital installations, exploring the intersection of art, technology, and human perception.

Key Takeaways

  • Creative domains like fashion, art, and design are significantly impacted by modern society and technology, making computer vision a crucial tool for innovation.
  • AI models, particularly diffusion models, offer powerful capabilities for generating diverse visual content, including images, videos, and dynamic 3D objects, pushing the boundaries of artistic expression.
  • Addressing biases in AI models and ensuring ethical considerations in creative applications of computer vision are critical challenges that require interdisciplinary collaboration.
  • The workshop emphasizes the importance of exploring new perspectives in computer vision, moving beyond traditional view-consistency to embrace view-inconsistency and create immersive, interactive, and anamorphic art forms.
  • Future research directions include improving controllability, speed, and robustness of generative models, as well as exploring opportunities at the intersection of 3D/4D and image/video models.

Methods / Models / Datasets Mentioned

  • stackGAN
  • txt2img
  • StyleGAN
  • DALL-E 3
  • Midjourney
  • Emu
  • Imagen
  • Firefly
  • Sora
  • L4GM
  • MVdream
  • Stable Diffusion
  • DreamFusion
  • Magic3D

Topics

Computer Vision · Fashion · Art · Design · Generative AI · Diffusion Models · 4D Content Creation · Bias in AI · Artistic Expression

Notes

Open for commentary — connections to other work, critiques, follow-up reading.