Problem
Detect manipulated or falsified media using semantic evidence that extends beyond low-level image artifacts.
Context
The project combined object-level cues, scene text, and human-pose signals across manipulated content.
My role
Developed computer vision and semantic reasoning pipeline components.
Constraints
- Evidence comes from multiple imperfect signals.
- Evaluation must account for diverse manipulations and failure modes.
TODO_REVIEW: Confirm public scope, datasets, collaborators, and NDA limits.
Architecture
The available source identifies object, text, and pose analysis feeding a semantic-forensics workflow.
TODO_REVIEW: Add the approved architecture and fusion strategy.
Technical decisions
- Treated semantic evidence as multiple complementary signals.
- Used detection and structured visual reasoning components.
Trade-offs
TODO_REVIEW: Document fusion, calibration, compute, dataset, and explainability
trade-offs.
Results
TODO_REVIEW: Add only approved, verifiable evaluation results.
Screenshots
TODO_REVIEW: Add approved diagrams or qualitative examples.
Related links
TODO_REVIEW: Add public program, paper, code, or demo links.