Video Analytics Research
Our research is aimed at building video analytics systems for industrial and security environments that understand context - recognizing people, equipment, and operational behavior in real time.
Compute & Acceleration
NVIDIA GPUs with CUDA-optimized pipelines for frame analysis, detection, and tracking
Accelerated inference (TensorRT / ONNX Runtime) and multi-camera scheduling to keep latency low
Algorithms for the Real World
Robust to dynamic lighting, glare, dust, and partial occlusions
Cross-camera association, zone logic, and event fusion for plant-wide awareness
Edge processing with smart buffering; optional cloud aggregation for fleet-level analytics
Outcomes
Intelligent surveillance and process monitoring, not just alarms
Actionable alerts, KPIs, and audit trails that improve safety, uptime, and decisions
XRPL for Secure Industrial Data
Using the XRPL as a cryptographic trust layer for industrial data-ensuring integrity, provenance, and controlled sharing across teams and partners.
Security Architecture
Off-ledger payload encryption (e.g., AES-GCM) with rotating keys; hashes anchored on XRPL for tamper-evident records
Account-based identities and multi-sign policies for machines, operators, and services
Time-stamped notarization of events, ML models, and firmware versions
Integration Pattern
Edge gateways sign and hash telemetry / video metadata → commit proofs to XRPL
Selective disclosure: share ciphertext and proofs with auditors or partners without exposing raw data
High throughput and low latency suitable for near-real-time pipelines
Outcomes
Verifiable data lineage from sensor to report
Simplified compliance and cross-organization collaboration
A foundation for secure automation, settlements, and audit-ready operations