modified some images, formatting and metadata

qdrant-landing/content/blog/qdrant-meets-twelve-labs-marengo-3.md

@@ -6,17 +6,16 @@ short_description: "Detect anomalies in live video by reframing surveillance as
 description: "Qdrant Edge, Twelve Labs Marengo 3.0, NVIDIA Metropolis VSS, and Vultr Cloud GPUs come together for real-time video anomaly detection. Detect what you've never seen before."
 preview_image: /blog/video-anomaly-detection-edge-to-cloud/preview.png
 social_preview_image: /blog/video-anomaly-detection-edge-to-cloud/preview.png
-date: 2026-03-16
+date: 2026-03-15
-author: Qdrant
+author: Thierry Damiba
 featured: false
 tags:
-  - news
   - blog
+  - anomaly detection
+  - Qdrant Edge
 ---
 
-![Sentinel dashboard: live camera grid with real-time anomaly alerts](/blog/video-anomaly-detection-edge-to-cloud/sentinel-dashboard.png)
+**What if your surveillance cameras could detect fights, accidents, intrusions, and equipment failures without ever being trained on those specific events?**
-
-What if your surveillance cameras could detect fights, accidents, intrusions, and equipment failures without ever being trained on those specific events?
 
 Traditional video classifiers need labeled examples of every anomaly type you want to catch. That breaks in the real world. You can't enumerate everything that could go wrong, and the moment something new happens, your model scores 0.0.
 
@@ -24,15 +23,16 @@ We built a system that takes a different approach: reframe anomaly detection as
 
 ![Reframing anomaly detection as vector search](/articles_data/video-anomaly-edge/vector-reframe.png)
 
-## The Idea
+### The Idea
 
 Index video embeddings of normal activity into Qdrant as a baseline. When a new clip arrives, embed it and search for its nearest neighbors. If the clip is far from anything in the baseline, it's anomalous. No anomaly labels required, no retraining when new anomaly types emerge.
 
-This works because the space of "normal" is learnable, but the space of "abnormal" is unbounded. A binary classifier trained on 13 crime categories will miss a forklift collision or a pipe burst. kNN distance from normal catches anything unusual by definition.
-
 ![kNN lookup: incoming clip embedded and searched against nearest neighbors in Qdrant](/articles_data/video-anomaly-edge/knn-lookup.png)
 
-## The Stack
+This works because the space of "normal" is learnable, but the space of "abnormal" is unbounded. A binary classifier trained on 13 crime categories will miss a forklift collision or a pipe burst. kNN distance from normal catches anything unusual by definition.
+
+
+### The Stack
 
 ![Tech stack overview: Qdrant Edge, Twelve Labs, NVIDIA VSS, Vultr](/articles_data/video-anomaly-edge/tech-stack-overview.png)
 
@@ -60,7 +60,7 @@ The system transforms live surveillance streams into:
 
 
 
-## Why Edge Matters
+### Why Edge Matters
 
 ![Escalation pipeline from edge to cloud](/articles_data/video-anomaly-edge/escalation-pipeline.png)
 
@@ -68,18 +68,18 @@ A 50-camera deployment generates 432,000 clips per day. Sending every clip to th
 
 The result is a system that scales with camera count without scaling cloud costs linearly.
 
-![Sentinel: real-time AI-powered anomaly detection powered by Qdrant Edge, Twelve Labs, Vultr, and NVIDIA](/blog/video-anomaly-detection-edge-to-cloud/sentinel-splash.png)
+### Build It Yourself
 
-## Build It Yourself
+<!-- ![Sentinel: real-time AI-powered anomaly detection powered by Qdrant Edge, Twelve Labs, Vultr, and NVIDIA](/blog/video-anomaly-detection-edge-to-cloud/sentinel-splash.png)  -->
 
 We published a full 3-part tutorial that walks through every component of this architecture with working code: from kNN anomaly detection theory through Qdrant Edge's two-shard design to baseline governance and Vultr deployment.
 
-- [Part I | Architecture, Twelve Labs, and NVIDIA VSS](/documentation/tutorials-build-essentials/video-anomaly-edge-part-1/)
+- [Part 1: Architecture, Twelve Labs, and NVIDIA VSS](/documentation/tutorials-build-essentials/video-anomaly-edge-part-1/)
-- [Part II | Edge-to-Cloud Pipeline](/documentation/tutorials-build-essentials/video-anomaly-edge-part-2/)
+- [Part 2: Edge-to-Cloud Pipeline](/documentation/tutorials-build-essentials/video-anomaly-edge-part-2/)
-- [Part III | Scoring, Governance, and Deployment](/documentation/tutorials-build-essentials/video-anomaly-edge-part-3/)
+- [Part 3: Scoring, Governance, and Deployment](/documentation/tutorials-build-essentials/video-anomaly-edge-part-3/)
 
 **GitHub**: [qdrant/video-anomaly-edge](https://github.com/qdrant/video-anomaly-edge)
 
-**Live Demo**: [qdrant-edge-video-anomaly.vercel.app](https://qdrant-edge-video-anomaly.vercel.app/)
+**Interactive Demo**: [qdrant-edge-video-anomaly.vercel.app](https://qdrant-edge-video-anomaly.vercel.app/)
 
 The concepts extend beyond surveillance. Manufacturing safety, retail analytics, traffic monitoring: anywhere you need to detect "something unusual" without defining every possible anomaly in advance.