Andrew Zisserman
University of Oxford
Top CV ResearchersFrontier Research MapScore: 9h-index: 1216,665 citations
Top CV Researcher — Rank #5 (top 10)
Professor of Computer Vision Engineering
Contributions
multiple view geometry, visual recognition, action understanding, self-supervised video learning
Why Selected
One of the principal architects of modern computer vision, especially geometry, recognition, and video understanding.
Score Breakdown
3
historical impact
2
recent visibility
2
current influence
2
asset availability
9
total
Frontier Research Map
Why Now
A concise senior-leader framing of how vision learns from temporal continuity, audio, and weak supervision rather than static labels alone.
Key Ideas
- -Video plus audio is a natural supervisory source for learning semantics and action.
- -Infant-like learning metaphors remain useful when thinking about scalable visual learning.
- -Recognition, sign understanding, retrieval, and description can be tied together by temporal multimodal training.
Open Questions
- ?How far can weakly supervised video carry semantic learning without curated labels?
- ?What forms of temporal structure do current video models still ignore?
- ?Can developmental analogies guide benchmark design rather than just motivate papers?
Canonical CV Leadershigh confidence
Sign in to access this content