# Edge Impulse

> Edge Impulse is the leading end-to-end edge AI MLOps platform, enabling developers and enterprises to build, deploy, and optimize embedded machine learning models for edge devices — from the smallest microcontrollers to gateways with the latest neural accelerators. Acquired by Qualcomm Technologies Inc. in March 2025, Edge Impulse continues to serve the entire hardware ecosystem under its original brand and mission.

Edge Impulse was founded in 2019 by Zach Shelby and Jan Jongboom with the mission of democratizing edge AI development. Headquartered in San Diego, CA (with offices in Santa Clara, CA), the platform covers the complete ML workflow: sensor data collection, feature engineering, model training and optimization, and deployment to any edge device that can run C++. The platform is SOC 2 Type II compliant and adheres to GDPR and CCPA standards.

## Key Benefits (BLERP framework)
- **Bandwidth** — Reduce data sent to the cloud by processing on-device
- **Latency** — Eliminate round-trip delays; inference runs locally
- **Economics** — Lower operational costs vs. cloud-dependent solutions
- **Reliability** — Reduced dependency on network connectivity
- **Privacy** — Data stays on the device and never leaves the edge

---

## Core Pages

- [Home](https://www.edgeimpulse.com): Overview of the Edge Impulse platform, value proposition, ecosystem partners, and primary calls to action.
- [Product](https://www.edgeimpulse.com/product): Full platform description covering sensor data management, feature engineering, model optimization (EON Suite), algorithms, and computer vision.
- [Pricing](https://www.edgeimpulse.com/pricing): Developer (free) and Enterprise (custom pricing) plan details, feature comparisons, compute limits, and licensing terms.
- [About](https://www.edgeimpulse.com/about): Company history, mission, founders (Zach Shelby & Jan Jongboom), Qualcomm acquisition (March 2025), values, and awards.
- [Security](https://www.edgeimpulse.com/security): SOC 2 Type II compliance, data protection controls, enterprise access management, and Vanta trust portal.
- [FAQs](https://www.edgeimpulse.com/faqs): Comprehensive answers covering platform capabilities, data formats, deployment options, pricing, Qualcomm acquisition details, and edge AI benefits.
- [Contact](https://www.edgeimpulse.com/contact): General inquiries, sales contact, and demo scheduling.

---

## Platform Capabilities

### Sensor Data Management
- Collect and visualize sensor, audio, and image data via flexible ingestion methods (UI, CLI, SDK, AWS S3, Google Cloud)
- Supported data formats: JPG, PNG, MP4, AVI, WAV, CBOR/JSON, CSV
- Monitor project performance over time with configurable dashboards
- Identify data quality issues with advanced analysis tools
- Automate data preparation through reusable cloud-based pipelines
- GenAI-assisted labeling with OpenAI and Hugging Face integration

### Feature Engineering
- State-of-the-art DSP algorithms for feature extraction optimized for sensor data
- Real-time on-device performance profiling (latency, RAM, flash)
- Feature importance rankings and DSP autotuning
- Custom signal processing blocks supported

### Model Optimization — EON™ Suite
- **EON Tuner**: AutoML tool that balances DSP algorithms and model architectures against device memory and latency constraints
- **EON Compiler**: Converts models to optimized C++ libraries — up to 70% less RAM, up to 40% less ROM vs. standard TensorFlow Lite
- Performance calibration measuring False Activation Rate and False Rejection Rate
- Support for TensorFlow, PyTorch, ONNX, Docker, and C++ export formats

### Specialized Algorithms
- **FOMO (Faster Objects More Objects)**: Real-time object detection for microcontrollers — 30× faster than MobileNet SSD, runs in <200 KB RAM
- **FOMO-AD**: Visual anomaly detection requiring no anomaly training data
- **YOLO-Pro**: Industrial-grade object detection in multiple sizes
- Anomaly detection across all sensor modalities
- Classification, regression, and object detection model types
- Multi-impulse deployment for combining multiple models

### Deployment Options
- C++ library (any device with C++ compiler)
- Arduino library
- WebAssembly
- Python library
- Docker container
- Mobile and desktop packages
- Supports up to 1,000 individual devices on the free Developer Plan

---

## Target Audiences

- [For AI Practitioners](https://www.edgeimpulse.com/ai-practitioners): Python SDK integration, DSP autotuning, EON Tuner AutoML, active learning, sensor-specific processing, and model profiling.
- [For Embedded Engineers](https://www.edgeimpulse.com/embedded-engineers): No-code ML, real-time on-device performance profiling, EON Compiler C++ generation, hardware-specific optimization, and AutoML.
- [For Product Leaders](https://www.edgeimpulse.com/product-leaders): Accelerated product development lifecycles, dashboard oversight, continuous learning models, SOC 2 compliance, and enterprise scalability.

---

## Computer Vision

- [Computer Vision Platform](https://www.edgeimpulse.com/computer-vision): Ultra low-power computer vision from industrial gateways to constrained cameras.
- [Visual Inspection](https://www.edgeimpulse.com/visual-inspection): Qualcomm Dragonwing™ Visual Inspection Suite — defect detection, OCR/label inspection, and assembly verification with no-code tools.

Key features: synthetic image generation (OpenAI, Hugging Face, NVIDIA Omniverse Replicator), GenAI-powered auto-labeling, Experiments for multi-model comparison, EON Tuner for vision architectures, FOMO / FOMO-AD / YOLO-Pro model family.

---

## Solutions by Industry

- [Agriculture](https://www.edgeimpulse.com/solutions/agriculture): Crop health monitoring, smart irrigation, remote farm monitoring, animal health, disaster protection, agricultural robotics.
- [Appliances](https://www.edgeimpulse.com/solutions/appliances): Smart water management, leak detection, predictive maintenance, anomaly detection via vibration, computer vision food classification.
- [Asset Tracking & Monitoring](https://www.edgeimpulse.com/solutions/asset-tracking-monitoring): Parcel tracking, smart animal wearables, remote environment monitoring, audio-based crime monitoring, cold chain logistics.
- [Buildings](https://www.edgeimpulse.com/solutions/buildings): Glass break detection, occupancy detection, HVAC predictive maintenance, smart environmental control, security cameras, voice control.
- [Environment](https://www.edgeimpulse.com/solutions/environment): Remote nature monitoring, wildfire detection, intelligent camera traps, fishery bycatch reduction, bioacoustics analysis.
- [Health](https://www.edgeimpulse.com/solutions/health): Heat exhaustion detection, fall detection, sleep quality monitoring, rural healthcare, baby monitoring, advanced medical imaging.
- [Human Interfaces](https://www.edgeimpulse.com/solutions/human-interfaces): Gesture control, wake word detection, head-tilt volume control, ML-powered coaching, acoustic event detection, smart home voice control.
- [Infrastructure](https://www.edgeimpulse.com/solutions/infrastructure): Grid power monitoring, water leak detection, crime monitoring, intelligent traffic control, smart street lights, road inspection.
- [Predictive Maintenance](https://www.edgeimpulse.com/solutions/predictive-maintenance): Water utility leak detection, appliance failure detection, assembly line inspection, equipment PdM, crops disaster protection, road defect detection.
- [Wearables](https://www.edgeimpulse.com/solutions/wearables): Wake word detection, gesture control, head-tilt control, ML-powered coaching, performance analysis, acoustic event detection.

---

## Case Studies

- [Revolutionizing Fall Detection with Edge AI](https://www.edgeimpulse.com/case-studies/revolutionizing-fall-detection-with-edge-ai): Elderly and mobility-limited fall detection using edge AI for personal safety.
- [HP: Bringing Voice Control to Earbuds & Headsets](https://www.edgeimpulse.com/case-studies/bringing-voice-control-to-earbuds-headsets): Always-on voice control for HP wireless headphones.
- [GlobalSense: Automotive Diagnostics with Edge AI](https://www.edgeimpulse.com/case-studies/globalsense-drives-new-standards-in-automotive-diagnostics-with-edge-ai): Advanced automotive event detection by San Diego startup GlobalSense.
- [Industrial Asset Tracking with NVIDIA Omniverse and TAO](https://www.edgeimpulse.com/case-studies/industrial-asset-tracking-with-nvidia-omniverse-and-tao): Pallet counting in global shipping and logistics using synthetic data.
- [Hyfe: Translating Coughs to Actionable Health Insights](https://www.edgeimpulse.com/case-studies/hyfe-translating-coughs-to-actionable-insights): Global cough audio datasets converted to health analytics.
- [Predictive Maintenance with Sony Spresense and CommonSense](https://www.edgeimpulse.com/case-studies/predictive-maintenance-with-sony-spresense-and-commonsense): Vibration and air-quality anomaly detection in industrial settings.
- [Adding AI to Optra Edge (Lexmark)](https://www.edgeimpulse.com/case-studies/adding-ai-to-optra-edge): ML integration into Lexmark industrial gateways.
- [Academia — Harvard SEAS Collaboration](https://www.edgeimpulse.com/case-studies/academia): Embedded ML research and education collaborative paper.
- [Smart Cold Chain Monitoring](https://www.edgeimpulse.com/case-studies/cold-chain): Real-time product lifecycle monitoring throughout shipment.
- [ML-Powered Agricultural Robotics](https://www.edgeimpulse.com/case-studies/ml-powered-agricultural-robotics): Precision irrigation and crop trial analytics.
- [Safer Roads, Powered by Machine Learning](https://www.edgeimpulse.com/case-studies/safer-roads-powered-by-machine-learning): Automatic wheel lug nut detection using on-device ML.
- [AI-Automated Hard-Hat Detection](https://www.edgeimpulse.com/case-studies/ai-automated-hard-hat-detection): Workplace safety enforcement through embedded ML.
- [IoT Gets an ML Boost, From Edge to Cloud](https://www.edgeimpulse.com/case-studies/iot-gets-an-ml-boost-from-edge-to-cloud): Raspberry Pi + Microsoft Azure IoT ML deployment.
- [Smart Grid Monitoring — RAM-1](https://www.edgeimpulse.com/case-studies/smart-grid-monitoring): World's smartest powerline monitoring system.
- [AI-Managed Crop Irrigation](https://www.edgeimpulse.com/case-studies/ai-managed-crops-irrigation): Arduino Edge Control + Edge Impulse maximize yield, minimize water waste.
- [SlateSafety: Predicting Heat Exhaustion](https://www.edgeimpulse.com/case-studies/slate-safety-predicting-heat-exhaustion): On-device real-time analytics protecting first responders from thermal stress.

---

## Key Partnerships & Integrations

- [NVIDIA Partnership](https://www.edgeimpulse.com/nvidia): NVIDIA Omniverse Replicator for synthetic datasets and digital twins; NVIDIA Jetson Orin support for production edge AI deployment.
  - [NVIDIA Omniverse Integration](https://www.edgeimpulse.com/nvidia/nvidia-omniverse)
  - [NVIDIA Jetson Orin Hardware](https://www.edgeimpulse.com/nvidia/nvidia-jetson-orin-hardware)
- [Qualcomm Partnership](https://www.edgeimpulse.com/qualcomm): Integration with Qualcomm AI Hub and 10+ Dragonwing-based devices (Rubik Pi 3, RB3 Gen 2, IQ9 EVK, Arduino UNO Q, and more); supports FOMO, YOLO-Pro, FOMO-AD; plans for VLM and multi-modal LLM support.
- [Arduino Integration](https://www.edgeimpulse.com/arduino-integrations): Native Arduino App Lab integration; deploy models directly to Arduino UNO Q and other Arduino boards.
- [Ecosystem Partners](https://www.edgeimpulse.com/ecosystem-partners): 50+ technology partners including AAEON, Advantech, Alif Semiconductor, Ambiq, AWS, BrainChip, CEVA, Microchip, Nordic Semiconductor, NXP, OnLogic, Particle, Renesas, Silicon Labs, STMicroelectronics, Syntiant, Wind River, and more.

---

## Plans & Pricing

| Feature | Developer (Free) | Enterprise (Custom) |
|---|---|---|
| Private projects | 3 | 10+ (configurable) |
| Public projects | Unlimited | Unlimited |
| Collaborators | Up to 3 | Unlimited |
| Compute time per job | 60 minutes | Unlimited |
| CPU memory | 16 GB | Configurable |
| GPU training | Yes | Yes |
| GenAI-assisted labeling | Yes | Yes |
| Full API access | Limited | Full |
| RBAC / SSO | No | Yes |
| Uptime SLA | — | 99.5% |
| Support | Forum & docs | Dedicated Slack/Teams + email |
| White-label | No | Available |
| Device deployments | Up to 1,000 | Configurable |
| Production use | Requires Enterprise | Yes |

- [ROI Calculator](https://www.edgeimpulse.com/roi-calculator)
- [Support Packages](https://www.edgeimpulse.com/support-packages)
- [Expert-Led Trial](https://www.edgeimpulse.com/expert-led-trial)
- [Buy vs. Build Analysis](https://www.edgeimpulse.com/buy-vs-build)

---

## Developer Resources

- Studio: https://studio.edgeimpulse.com
- Documentation: https://docs.edgeimpulse.com/docs
- Forum: https://forum.edgeimpulse.com
- [Projects Gallery](https://www.edgeimpulse.com/projects/all)
- [Content Library](https://www.edgeimpulse.com/content-library)
- [Academic Resources](https://www.edgeimpulse.com/academic-resources)
- [University Program](https://www.edgeimpulse.com/university)
- [Schedule a Demo](https://www.edgeimpulse.com/schedule-a-demo)

---

## eBooks & Guides

- [5 Rising Trends for AI Adoption in Manufacturing](https://www.edgeimpulse.com/ebooks/5-rising-trends-for-ai-adoption-in-manufacturing)
- [AI at the Edge](https://www.edgeimpulse.com/ebooks/ai-at-the-edge-book-mail)
- [Building the Business Case for Edge AI](https://www.edgeimpulse.com/ebooks/building-the-business-case-for-edge-ai)
- [Powering Smart Manufacturing with Edge AI](https://www.edgeimpulse.com/ebooks/powering-smart-manufacturing-with-edge-ai-guide)
- [The Build vs. Buy Reckoning](https://www.edgeimpulse.com/ebooks/the-build-vs-buy-reckoning)
- [The Convergence of Edge AI and Cloud](https://www.edgeimpulse.com/ebooks/the-convergence-of-edge-al-and-cloud)
- [Wevolver Guide to Generative AI at the Edge](https://www.edgeimpulse.com/ebooks/wevolver-guide-to-generative-ai-at-the-edge)

---

## Events

- [Events Hub](https://www.edgeimpulse.com/events)
- [Imagine Conference (annual flagship event)](https://www.edgeimpulse.com/imagine)
- [Imagine 2026 Speakers](https://www.edgeimpulse.com/imagine-2026-speakers/jan-jongboom)
- Past webinars: GenAI for the Edge, Edge ML from Wrist to Factory, Synthetic Data with NVIDIA Omniverse, Industrial Acoustic/Vibration Monitoring, Computer Vision at the Edge, Visual Inspection & Anomaly Detection, Anomaly Detection in PLCs.

---

## Legal & Compliance

- [Privacy Policy](https://www.edgeimpulse.com/legal/privacy-policy)
- [Terms of Service](https://www.edgeimpulse.com/legal/terms-of-service)
- [Enterprise Terms of Service](https://www.edgeimpulse.com/legal/enterprise-terms-of-service)
- [Responsible AI License](https://www.edgeimpulse.com/legal/responsible-ai-license)
- [Accessibility Statement](https://www.edgeimpulse.com/legal/accessibility-statement-for-edge-impulse)
- [DMCA Policy](https://www.edgeimpulse.com/legal/dmca-digital-millennium-copyright-act-policy)