Paper Reading Session: Localizing Object-level Shape Variations with Text-to-Image Diffusion Models
About Event:
Join us for an engaging paper reading session on "Localizing Object-level Shape Variations with Text-to-Image Diffusion Models," led by Rishubh Parihar, a PhD candidate specializing in Generative Modeling at IISc and an IIT Delhi 2018 graduate. The paper explores techniques to generate images showcasing variations in object shapes, allowing for detailed object-level shape exploration. Discover how text-to-image models handle shape variations while maintaining semantic coherence.
Paper Link:
https://arxiv.org/abs/2303.11306
Paper Description:
Text-to-image models initiate workflows that start with exploration, hindering users from focusing on specific objects. This paper introduces techniques enabling the generation of image collections depicting variations in the shape of specific objects, enabling object-level shape exploration. It addresses challenges in creating plausible shape variations while controlling object semantics and presents localization techniques leveraging self-attention and cross-attention layers.
Workshop Agenda:
Guided Reading: Rishubh will lead a step-by-step exploration of the paper's sections.
Discussions: Engage in open discussions to analyze and critique the paper's methodologies.
Summarization: Rishubh will conclude by summarizing the key points discussed during the session.
Target Audience:
Students, researchers, and enthusiasts interested in generative modeling, computer vision, or AI-related fields.
Workshop Prerequisites:
Participants are encouraged to read the paper beforehand to engage actively during the session. A basic understanding of machine learning concepts and familiarity with generative models will be beneficial.
https://arxiv.org/abs/2303.11306
Application Process:
Limited seats available. Apply as soon as possible to secure your spot.
Query: Please feel free to write to
Arpit https://wa.me/9971677857
Date & Time:
February 11th, 11 am IST
Venue:
Online via Video Call