Imaging Science Thesis Defense: Towards Human-Embodied Visual Intelligence
Imaging Science Thesis Defense
Towards Human-Embodied Visual Intelligence
Cheng Han
Imaging Science
Rochester Institute of Technology
Register for Zoom Link Here
Abstract:
The rapid development of artificial intelligence (AI) in the domain of computer vision has garnered considerable attention across diverse research fields. However, this progress is met with an array of challenges that hinder them from real-world deployments. Specifically, three major challenges will be discussed. i) In current visual models, model designs are often specialized for targeted tasks, thus constraining the models’ generalizability; ii) The majority of these models primarily optimize performance, rendering them less effective for decision trustworthy and interpretability; iii) The ongoing pursuit of superior performance has led to the scaling up of visual models, resulting in substantial computational costs. While these challenges persist, human visual intelligence, on the other hand, naturally navigates these challenges. I thus seek to embody and mimic the human’s capabilities via three solutions: Universal Visual Learners, Generalizable, Interpretable Visual Intelligence, and Carbon-efficient Visual Intelligence Systems.
Intended Audience:
All are Welcome!
To request an interpreter, please visit myaccess.rit.edu
Event Snapshot
When and Where
Who
This is an RIT Only Event
Interpreter Requested?
No