Edge Impulse

Edge Impulse is a web-based edge AI development platform for developers, embedded engineers, and ML teams building, training, and deploying machine learning models on resource-constrained devices.

Key capabilities include:

• Data acquisition and dataset management from real devices and real-time sources, including device and phone-based collection, cloud storage (for example, Amazon S3), and synthetic data generation
• End-to-end model workflows in a graphical UI and via an open web API, covering data exploration, processing blocks, feature extraction, training, hyperparameter tuning, and on-device performance profiling
• Edge-focused model optimization with the EON Compiler, targeting lower memory footprints (up to 70% less RAM and 40% less flash, per provided brief) and faster inference for neural networks and classical ML
• Deployment to diverse hardware (MCUs, NPUs, CPUs, GPUs, gateways, sensors, cameras), exporting artifacts such as C++ libraries, NVIDIA TensorRT, and Linux processes

Pricing: Developer (Free) and Enterprise (Paid) plans.

Notable for its hardware-agnostic deployment formats and practical tooling for taking edge ML from data collection to on-device testing across multiple tasks, including classification, anomaly detection, regression, image/audio transfer learning, and object detection.