AD Research Hub
DashboardPeoplePapersWorkshopsDatasetsResearch Map
Login
⌘K
AD Research Hub — Anomaly Detection in Computer Vision
← Back to People

Sanja Fidler

NVIDIA and University of Toronto

Top CV ResearchersFrontier Research MapScore: 9h-index: 1122 citations
HomepageNVIDIA research profileNeurIPS 2025 keynote pageUniversity of Toronto homepageSemantic Scholar
Top CV Researcher — Rank #10 (top 10)

Vice President of AI Research at NVIDIA; associate professor

Contributions

3D vision, spatial intelligence, world models, autonomous driving, multimodal representations

Why Selected

One of the clearest current leaders in 3D vision, world models, and spatial intelligence, with strong academic and industrial influence.

Score Breakdown

2

historical impact

3

recent visibility

2

current influence

2

asset availability

9

total

Frontier Research Map

Featured Work

GEN3C: 3D-Informed World-Consistent Video Generation with Precise Camera Control

official project page — 2025-02-01

Why Now

A concrete example of the field moving from image realism to controllable, geometrically consistent video worlds.

Key Ideas

  • -World-consistent video generation increasingly depends on explicit 3D state rather than purely latent video tokens.
  • -Camera control is an acid test for whether a model has internalized scene geometry.
  • -Generative video and perception are converging around shared 3D abstractions.

Open Questions

  • ?Will world models be trained primarily for simulation, perception, or generation?
  • ?How much explicit geometry should remain in the loop as models scale?
  • ?What benchmarks capture consistency across long horizons and interventions?
Canonical CV Leadershigh confidence
Cross-References

Themes

world models3D consistencyvideo generationvideo

Sign in to access this content

Sign in