Case Study

A Video Pipeline that Strips PII On-Device Before Anything Streams

The client can extract value from video while privacy is enforced by architecture, before data ever leaves the box, rather than trusting a downstream filter.

Industry

Edge & Offline AI

Services We offer

Edge AI

Computer Vision

Privacy Engineering

Streaming

About the client

A medical-technology organization handling sensitive patient video.

Problem Statement

Patient video had to be processed for insight while meeting strict privacy and compliance rules. That meant handling many input types and removing PII before anything left the device.

The Challenges

Redact before it leaves

PII removal runs on the edge in the live path, so raw identifiable frames never transit the network.

Many sources, one pipeline

A modular design absorbs diverse capture hardware and formats without rewriting the core.

Approach

We built a modular, edge-first pipeline:

Flexible video input handling, including capture cards for clinical sources
PII removal performed on the edge
A modular processing pipeline with real-time streaming
Storage, retrieval, monitoring and logging around it

The Solution

Frames are captured, scrubbed of PII on-device, and only then streamed or stored. The pipeline is assembled from OpenVINO inference and a GStreamer media path, with ZeroMQ and an MQTT broker moving data between modular stages.

Read more case studies

An Offline AI Decision Register for the Operating Theatre, Zero Data Egress

Learn More

An End-to-End Model Preparation and Validation Pipeline

Learn More

Unified Device Management for Hospital Networks with Azure Arc

Learn More

Get Detailed Case Study

A Video Pipeline that Strips PII On-Device Before Anything Streams

A modular video pipeline that ingests diverse feeds, removes personally identifiable information on the edge, and streams in real time, built for strict privacy and compliance.

Apportez-nous le cas limite que votre cloud ne peut pas atteindre. Concevons-le ensemble.