FP-AI-VISION1 is an STM32Cube function pack featuring examples of computer vision applications based on a Convolutional Neural Network (CNN).
FP-AI-VISION1 is composed of software components generated by the X-CUBE-AI Expansion Package complemented with application software components dedicated to the AI-based computer vision application.
The application examples provided in the function pack are food recognition (recognizing among 18 classes of common food), person presence detection (identifying whether a person is present in the image or not), and people counting (counting the number of persons in a scene) based on an object detection Neural Network model.
FP-AI-VISION1 implements an advanced computer vision application using STM32_AI_Runtime Neural Networks libraries. Libraries are based on pretrained models and are generated with the X-CUBE-AI Expansion Package for the STM32CubeMX tool.
The function pack demonstrates the integration of two types of Neural Network model: 32-bit floating-point model and 8-bit quantized model. It also demonstrates model integration in different memory configurations (relying only on MCU internal memory or using also external memories).
The FP-AI-VISION1 function pack includes a feature-rich image processing library, STM32_ImageProcessing_Library, that enables both the common processing tasks (such as image rescaling and pixel color conversion) and advanced processing tasks (such as face detection). The STM32_ImageProcessing_Library also provides a set of utility functions to read and write several file formats.
The FP-AI-VISION1 function pack also includes the drivers for the camera as well as the framework for capturing images into the frame buffer, preprocessing the content of the frame buffer, and running the Neural Network inference.
The FP-AI-VISION1 function pack features a USB webcam application, which can be used to create image and video datasets as well as to perform live testing on the host. The FP-AI-VISION1 runs on the STM32H747I-DISCO Discovery board connected to the B-CAMS-OMV camera module bundle (advised) or STM32F4DIS-CAM camera daughterboard (legacy only).