Skip to content
VirtualOnlinePast event

Best of WACV 2026: Live Research Presentations

Event artwork for Best of WACV 2026: Live Research Presentations

This event has ended

This listing is kept as an archive. Explore upcoming events to find your next one.

See upcoming events

Supporting this event

Sponsors

Hugging Face
About this event

What to expect

Get a front-row seat to cutting-edge computer vision research without leaving your desk. This online session brings you the best talks from WACV (Winter Conference on Applications of Computer Vision) 2026, streamed live from the conference authors themselves. Hosted by the Silicon Valley AI Innovators Meetups, this 2-hour virtual event showcases three groundbreaking research presentations that are pushing the boundaries of AI and machine learning.

Featured Research Presentations

Zero-Shot Coreset Selection via Iterative Subspace Sampling presented by Brent Griffin, Principal Machine Learning Scientist at Voxel51. Learn how ZCore tackles deep learning's massive dataset problem through a zero-shot approach that works without labels or prior training. The research demonstrates outperforming previous label-based methods on ImageNet at a 90% prune rate, eliminating the need to annotate over one million images — a game-changer for anyone dealing with large-scale training costs.

ENCORE: A Neural Collapse Perspective on Out-of-Distribution Detection by A.Q.M. Sazzad Sayyed, Ph.D. candidate at Northeastern University. Discover a post-hoc method for detecting out-of-distribution inputs that leverages neural collapse geometry. This lightweight, training-free approach enhances uncertainty estimation across architectures including CNNs and vision transformers, with critical implications for safety-critical and edge AI systems.

Synthesizing Compositional Videos from Text Description presented by Shanmuganathan Raman, Professor at IIT Gandhinagar. Explore Video-ASTAR, a training-free approach addressing the compositional challenges in text-to-video generation. The research introduces novel techniques for multi-object scene composition and coherent multi-action video generation through token-swapping and latent interpolation.

The Perceptual Observatory: Characterizing Robustness and Grounding in MLLMs by Fenil Bardoliya from Arizona State University's CORAL Lab. Examine whether multimodal large language models truly understand what they see through interpretable tasks across pixel-level corruptions and diffusion-based stylized illusions. The research reveals that scaling language

Running an AI event?

List it here, free.

Meetups, workshops and conferences go in front of thousands of builders after a quick review.

List your event