STMicroelectronics Doubles STM32 AI Model Zoo to 140+ Models with Native PyTorch Support for Edge AI

STMicroelectronics has released version 4.0 of the STM32 AI Model Zoo, expanding the number of model families from 30 to 60 and providing more than 140 ready-made models for vision, audio, and sensing applications on STM32 microcontrollers.

The updated Model Zoo now includes native support for PyTorch models in addition to existing TensorFlow Lite, Keras, LiteRT, and ONNX formats. It incorporates compressed and sub-byte quantized models and provides scripts for training and integration with application libraries.

The STM32 AI Model Zoo is hosted as an open-source standalone repository at https://github.com/STMicroelectronics/stm32ai-modelzoo/ and forms part of the ST Edge AI Suite.

The release targets embedded AI development for STM32-based devices including wearables, smart cameras, sensors, security systems, and robotics. ST states that its AI tools currently support over 160,000 projects per year.

The STM32 family is used in consumer appliances, wearables, communication infrastructure, smart grids, smart cities, industrial automation, and low-earth-orbit satellites. The company also offers AI-accelerated MCUs in the STM32N6 series.