Cyber-Servalence
An intelligent, hardened physical security and access monitoring architecture integrating real-time OpenCV facial recognition, low-latency MJPEG video pipelines, and Fernet-encrypted telemetry storage.
System Architecture & Processing Pipeline
Facial Detection & Embeddings
Utilizes OpenCV Haar cascades and deep convolutional face embeddings to extract facial feature vectors from video frames, comparing them against authorized personnel profiles using cosine similarity thresholds.
Low-Latency MJPEG Streaming
Asynchronous camera worker threads capture raw video feeds and stream multi-part JPEG responses over HTTP, enabling real-time monitoring on standard web browsers without heavy client plugins.
Fernet Symmetric Encryption
Personnel database records, identification logs, and camera telemetry are secured at rest using Fernet 128-bit AES-CBC encryption with PKCS7 padding and HMAC authentication.
JWT Role-Based Access Control
Operator sessions are authenticated through signed JSON Web Tokens (JWT) with configurable expiration windows, preventing unauthorized access to live camera feeds or sensitive event logs.