AN14718: USB Camera Object Detection Pipeline on i.MX RT700

Overview

This document describes the design and implementation of a real-time object detection pipeline on the NXP i.MX RT700 crossover microcontroller. It explains how to leverage the device's heterogeneous architecture, including its dual Cortex-M33 cores, Neural Processing Unit (NPU), Vector Graphics Processing Unit (VGPU), and hardware JPEG decoder (JPEGDEC). The note compares a single-core baseline against a dual-core optimization that offloads image capture and preprocessing to the sense domain (CPU1) while reserving the compute domain (CPU0) for inference. The implementation uses the FastestDet model, quantized to INT8 and optimized via the eIQ Toolkit, to achieve efficient edge AI performance for monitoring tasks.

Referenced Parts