Assessing the Efficacy of TinyML Implementations on STM32 Microcontrollers: A Performance Evaluation Study
Authors:
- 11 July 2024
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 1, 267-271
Abstract
In recent years, there has been a growing need for efficient deployment of deep learning models on resource-limited edge devices. This trend has prompted the development of NVIDIA’s TAO Toolkit and TensorFlow Lite Micro framework as promising solutions for running deep-learning models on STM32 microcontrollers. This paper presents a comparative study between these two tools, aiming to evaluate their performance, resource usage, and ease of deployment. By delving into their respective strengths and limitations, we seek to provide insights into the best practices for optimizing deep learning models in edge computing scenarios. Through empirical analysis, we assess the impact of model optimization techniques on classification accuracy, memory usage, and computational efficiency. Our findings reveal trade-offs between model complexity and resource consumption, shedding light on the strengths and limitations of each tool. Additionally, we explore the feasibility of deploying optimized models on STM32 microcontrollers using the STM32Cube.AI Developer Cloud platform. Insights from this study contribute to the advancement of efficient edge AI solutions by providing guidance on selecting appropriate optimization tools for specific deployment scenarios.Keywords
This publication has 16 references indexed in Scilit:
- Tiny Machine Learning Business Intelligence in the Semiconductor Industry: A Case Study2023
- TinyissimoYOLO: A Quantized, Low-Memory Footprint, TinyML Object Detection Network for Low Power Microcontrollers2023
- Methods for Pruning Deep Neural NetworksIEEE Access, 2022
- Design of an Embedded Machine Learning Based System for an Environmental-friendly Crop Prediction Using a Sustainable Soil Fertility Management2021
- An SRAM Optimized Approach for Constant Memory Consumption and Ultra-fast Execution of ML Classifiers on TinyML Hardware2021
- Fast and Accurate Inference on Microcontrollers With Boosted Cooperative Convolutional Neural Networks (BC-Net)IEEE Transactions on Circuits and Systems I: Regular Papers, 2020
- Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inference2018
- Deep Learning with Low Precision by Half-Wave Gaussian Quantization2017
- Structured Pruning of Deep Convolutional Neural NetworksACM Journal on Emerging Technologies in Computing Systems, 2017
- Model compression2006