PROPOSEDPOD 02: Edge AI & Vision Systems
Multi-Camera Video Stream Anomaly Detection Pipeline
Objective & Abstract
Designing a zero-copy frame ingestion buffer for temporal anomaly localization in campus perimeter feeds without CPU memory serialization bottleneck.
Investigated Methodologies
- Zero-copy memory mapping
- Temporal bounding box tracking
- RTSP stream ingestion
Datasets & Benchmarking Corpora
Avenue Dataset
Synthetic Perimeter Streams
All training and evaluation datasets undergo ethics audit and licensing compliance per AIAC Research Council standards.