/Tengine-lite

An extremely lite inference engine, written in C. Cortex-M and RTOS are supported as well

Primary LanguageCApache License 2.0Apache-2.0

Tengine-lite

Tengine-lite is lighten version of Tengine for Cortex-M & RTOS, written in C.

Supported Ops

  • Convolution
  • Pooling
  • Fully-connected
  • Relu
  • Softmax

Example

Cortex-M-KWS is an Keyword Spotting(KWS) demo on Cortex-M(STM32F7). It takes an quantized tensorflow model, converts it into .c and .h files, and runs the model with Tengine-lite. For more details, please visit this project Cortex-M-KWS.