majianjia/nnom

Health v2 · maintenance only

Adoption not part of the score

1164 stars · 281 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

NNoM is a high-level neural network inference library written in C for microcontrollers. It converts Keras models into optimized on-device models with one line of code, supporting complex architectures like ResNet and LSTM with per-channel quantization.

Use cases

• run neural network inference on a microcontroller
• deploy a keras model to an embedded mcu
• do tinyml keyword spotting on an arm cortex-m
• run cnn image classification on an iot sensor node
• quantize a model for low-memory embedded inference
• evaluate model accuracy onboard with confusion matrix and top-k

When to choose

• you need to run neural network inference directly on a microcontroller in C
• you want a one-line conversion from Keras models to embedded inference code
• you need support for complex architectures like ResNet, Inception, or LSTM on MCUs
• you want onboard evaluation tools like runtime analysis and confusion matrix

When to avoid

• you need to train models on-device rather than only run inference
• your target is a full OS device like a phone or server where TensorFlow Lite or PyTorch is more appropriate
• you need a framework with broad community tooling beyond Keras model conversion

Facets

library · maturity stable

machine-learning deep-learning embedded machine-learning embedded-systems iot deep-learning embedded c iot neural-network-inference microcontroller keras-deployment quantization tinyml mcu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="majianjia/nnom")