Introducing Open Edge Platform 2025.1 — Enhanced AI Tools & Industry-Focused Applications at the Edge

[biapalmeiro]

Hi everyone 👋,

We’re thrilled to announce the Open Edge Platform 2025.1 release, packed with exciting enhancements and industry-specific sample applications designed to accelerate your edge AI development. Below is a quick overview of the highlights—let's discuss!

---

Key Highlights:

1. Industry Sample Applications

• Metro AI Suite – Smart Traffic Intersection
  Real-time multi-camera analytics + sensor fusion (LiDAR/radar) for advanced traffic management (Video, Spatial AI).

• Gen AI Capabilities

  • Video Search Q&A: Query video content with Gen AI.
  • Smart NVR: AI-powered Network Video Recorder for smarter insights from video feeds.

• Manufacturing AI Suite – Wind Turbine Predictive Maintenance
  Handles time series ingestion and anomaly detection—integrated with MQTT, OPC-UA, OPC data streams.

• Retail AI Suite

  • Order Accuracy: DL Streamer + OpenVINO pipelines for detecting order errors in restaurant settings.
  • Loss Prevention: Shopper behavior and scene analytics to reduce shrinkage.

---

2. Feature Enhancements

Geti™: Supports fine-tuning CV models on Intel Arc GPUs. Adds pose estimation, configurable model sizing, and optimized data augmentation.

DL Streamer Framework: Now Windows-compatible for fast media analytics at the edge.

DL Streamer Pipeline Server: Scalable microservice for up to 4 cameras—real-time counting, tracking, and detection.

OpenVINO™ Upgrades:

GGUF reader integration (support for llama.cpp models)

Optimized LoRA adapters (LLM, VLM, text-to-image pipelines)

Support for image-to-image and inpainting transformer pipelines (Flux.1, Stable Diffusion 3)

---

3. Platform Enhancements: Streamlining Remote Management & Resource Utilization

The 2025.1 release takes a major step forward in simplifying remote management and optimizing resource utilization for edge environments.

• Edge Manageability Framework: Now integrates Intel® vPro™ and OpenAMT, enabling true out-of-band hardware management for vPro-capable platforms. This includes remote power on/off and reboot capabilities without host dependency, giving operations teams seamless control of geographically distributed edge nodes—reducing the need for on-site intervention.
• Edge Microvisor Toolkit: Adds discoverability of Intel® Arc™ B-Series Graphics and other components to containerized apps and VMs, unlocking more compute power for distributed AI workloads. The improved under-the-hood fine-tuning minimizes latency, enhances data security, and improves interoperability, resulting in better-performing AI applications at the edge.

This combination of near zero-touch onboarding, hardware-level optimization, and secure scalability makes the Open Edge Platform not just a toolkit for building applications but a complete solution for running them optimally on Intel silicon.