Back to browse
GitHub Repository

Foundational framework for mission-critical surveillance, autonomous video intelligence, and situational awareness.

4 starsPython

Sentinel Core – Open-source AI video search with real-time dashboard

by ab-abg·Feb 16, 2026·1 point·0 comments

AI Analysis

●●SolidWizardryNiche Gem
The Take

Combines YOLOv8 detection, CLIP text-to-video embeddings and FAISS indexing so you can query live and archived camera feeds with plain English; the FastAPI + WebSocket dashboard for multi-RTSP streams and zone-based alerts shows practical, deployment-minded thinking. They also call out Apple MPS/CUDA paths and a local SQLite 'sovereign memory', which makes this useful for air-gapped or edge setups — I'd still want to see performance benchmarks, scale testing, and a clear stance on privacy/ethics before recommending it for sensitive deployments.

Category
Target Audience

Security engineers, SREs/DevOps, integrators building surveillance/physical AI systems, and privacy-conscious orgs wanting local AI video search

Post Description

We just released a major update to Sentinel Core, our open-source AI-powered video intelligence engine.

The new release includes a real-time multi-camera dashboard, semantic video search, and an upgraded detection pipeline.

Sentinel Core is built around an AI-native architecture:

- YOLOv8 for real-time object detection

- CLIP for semantic search (text-to-video queries)

- FAISS for vector indexing

- FastAPI backend with WebSocket streaming

- Fully air-gapped deployment capable

You can now:

- Stream multiple RTSP/IP camera feeds in a live grid view

- Search footage using natural language (e.g. “blue car near gate between 2-5pm”)

- Perform cross-camera semantic retrieval

- Configure zone-based alerts

- Run everything locally without cloud dependency

We designed this as a foundational engine for modern surveillance and physical AI systems.

Would appreciate feedback on:

- Architecture decisions

- Performance optimizations

- Semantic search accuracy

- Deployment improvements

Happy to answer technical questions.

Similar Projects

AI/ML●●Solid

Sentinel Core – Open-source AI video search and detection engine

They stitched together sensible primitives — YOLO for real‑time detection and CLIP for natural‑language search — and packaged it for truly local deployments (SQLite "sovereign memory", MPS/CUDA hints, RTSP ingest). The README and demo show a coherent pipeline and multi‑camera focus, which is rarer than you'd think; what I'd like to see next is clearer install docs, reproducible benchmarks, and more evidence of robustness under production loads.

WizardryNiche Gem
ab-abg
123mo ago
Open Source●●Solid

Controlling an air-gapped vacuum from Home Assistant via text to speech

It gets control of an offline robot by doing something almost obvious in hindsight: play the robot's own wake-word + command audio from a nearby Pi, triggered by Home Assistant POSTs. That avoids rooting/Valetudo and keeps firmware intact, while the repo packages Docker, Ansible and embedded Piper TTS for a near plug-and-play hack — though reliability hinges on speaker placement, timing and device-specific command sets.

Niche GemBig Brain
nbr23
103mo ago