Hi!
Self-nominating microflow-rs, a robust and efficient TinyML inference engine for embedded systems.
With MicroFlow, it’s possible to locally run AI models on embedded systems. MicroFlow uses a compiler-based approach, making it possible to efficiently use the memory by pre-processing the model and statically allocating on the stack only.
MicroFlow has been tested with multiple MCUs and model architectures, including a speech command recognizer (TinyConv) on an 8-bit ATmega328 (Arduino Uno), a person detector (MobileNet v1) on an ESP32, and more.