Hardware Guide

STM32U5 for People Counting with TensorFlow Lite Micro

STMicroelectronics's STM32U5 is a solid choice for people counting using TFLite Micro. The cortex-m33 core at 160 MHz with 786 KB SRAM accommodates 200 KB models with room for application logic. DSP extensions available.

Hardware Specs

Spec STM32U5
Processor ARM Cortex-M33 @ 160 MHz
SRAM 786 KB
Flash 2 MB
Key Features Ultra-low-power (best-in-class Cortex-M33), TrustZone hardware security, Hardware crypto (AES/PKA/HASH), SMPS for power efficiency, Up to 2514 KB SRAM on U5A5/U5G9 variants
Connectivity USB OTG HS
Price Range $6 - $15 (chip), $20 - $50 (dev board)

Compatibility: Good

The STM32U5's 786 KB SRAM provides 4.1x the 192 KB minimum for people counting. This generous headroom means the 200 KB model tensor arena, sensor input buffers, and application logic (camera polling, USB OTG HS stack, state management) all fit without contention. The remaining 286 KB after model allocation supports complex application features. The STM32U5 provides 2 MB of flash memory, which accommodates the TFLite Micro runtime and 200 KB model. Space remains for firmware and basic OTA capability. The STM32U5 combines Cortex-M33 with TrustZone for secure ML inference and ultra-low power consumption. Its 786 KB SRAM is among the largest in low-power MCU families. The SMPS voltage regulator extends battery life in duty-cycled inference scenarios. People Counting requires camera input. The STM32U5 lacks native peripheral support for some of these sensors, requiring external interface circuitry. A camera interface (DVP/DCMI) is not available — SPI-based camera modules may work but with reduced frame rates. Evaluate whether the peripheral gap justifies an alternative MCU with native support. TFLite Micro's static memory allocation model maps well to the STM32U5's memory architecture — define a fixed tensor arena at compile time with no runtime heap fragmentation risk. The framework's operator coverage supports convolutional, depthwise-separable, and pooling layers needed for people counting. Model conversion uses the standard TFLite converter with int8 post-training quantization. At $6-15 per chip ($20-50 for dev boards), the STM32U5 is a reasonable investment for people counting deployments. Key STM32U5 features for this workload: Ultra-low-power (best-in-class Cortex-M33), TrustZone hardware security, Hardware crypto (AES/PKA/HASH), SMPS for power efficiency, Up to 2514 KB SRAM on U5A5/U5G9 variants.

Getting Started

  1. 1

    Set up STM32U5 development environment

    Install STM32CubeIDE with the latest STM32Cube firmware package. Create a project targeting the STM32U5 and verify basic functionality (blink LED, serial output). For TFLite Micro, clone the framework repository and add it as a library dependency. Ensure the toolchain supports C++11 or later for the ML runtime.

  2. 2

    Collect camera training data

    Connect a camera module (e.g., OV2640 via DVP/SPI) to the STM32U5. Write a data logging sketch that captures camera readings at the target sample rate and outputs via serial/SD card. Collect 1000+ labeled samples across all classes. Capture images at the model input resolution (96×96 or lower).

  3. 3

    Train and quantize model for TFLite Micro

    Build a quantized MobileNet-SSD or YOLO-Tiny in TensorFlow or PyTorch. Apply int8 post-training quantization — this typically reduces model size by 4x with minimal accuracy loss. Convert to .tflite and generate a C array (xxd -i model.tflite > model_data.h). Target model size: under 200 KB to fit the STM32U5's 786 KB SRAM with room for application code.

  4. 4

    Deploy and validate on STM32U5

    Include the TFLite Micro runtime and compiled model in your STMicroelectronics project. Allocate a tensor arena of 300-500 KB in a static buffer. Run inference on live camera data and compare predictions against your test set. Log results to serial for desktop validation. Measure inference latency and peak RAM usage to verify they meet application requirements.

Alternatives

Explore More

FAQ

What camera resolution works for people counting on STM32U5?
On-device people counting models typically use 96×96 or 128×128 pixel grayscale input. The STM32U5's 786 KB SRAM constrains this: a 96×96 grayscale frame is ~9 KB, while 128×128 RGB would need ~49 KB. Without a native camera interface, use an SPI camera module (e.g., ArduCAM Mini) with reduced frame rates. Always downsample in firmware before inference.
How do I update the people counting model on STM32U5 in production?
Without wireless connectivity, model updates require physical access via USB/JTAG. For field deployments, consider adding a wireless module or using an MCU with built-in connectivity. Always validate model integrity with a checksum before switching to the new version.
What size people counting model fits on STM32U5?
The STM32U5 has 786 KB SRAM and 2 MB flash. A typical people counting model is 200 KB after int8 quantization. The tensor arena needs 300-400 KB at runtime. After model allocation, approximately 386 KB remains for application logic, sensor drivers, and USB OTG HS stack.

Build Vision AI Pipelines in ForestHub

Connect cameras to on-device inference — design detection workflows visually and compile to optimized firmware.

Get Started Free