Battery Optimized People Counting Using FIR and AI
Explore battery-optimized people counting using FIR and AI, covering product development, power efficiency challenges, AI implementation, and quality assurance for innovative crowd monitoring solutions.
TinyML Talks Germany - AutoFlow - An Open Source Framework to Automatically Implement Neural
Explore AutoFlow: an open-source framework for automating neural network implementation on embedded devices, covering model building, optimization, and deployment.
TinyML Talks - Oculi Is Putting the Human Eye in A.I.
Innovative imaging technology mimicking human vision for efficient machine perception, combining sensing and pre-processing on a single chip for superior energy-delay performance in AI applications.
Lightweight Face Detection for Ultra-Low Power AI Processors - tinyML Asia 2021
Lightweight face detection method for ultra-low power AI processors, featuring real-time application and accuracy in various scenarios. Explores implementation on Himax WE-I Plus platform for always-on sensing products.
An Approach to Dynamically Integrate Heterogeneous AI Components in Multimodal User Authentication - tinyML Asia 2021
Approach to dynamically integrate heterogeneous AI components for multimodal user authentication, focusing on flexibility, performance, and energy efficiency in battery-powered devices.
On-Device Neural End-to-End Speech Recognition and Synthesis Algorithms - A Review
Comprehensive review of on-device neural speech recognition and synthesis algorithms, exploring end-to-end models, optimization techniques, and their potential for on-device implementation.
Putting AI on a Diet: TinyML and Efficient Deep Learning
Efficient deep learning techniques for deploying AI on resource-constrained devices, including model compression, neural architecture search, and new design primitives for various applications.
AIfES - An Open-Source Standalone AI Framework for Almost Any Hardware
Explore AIfES, an open-source AI framework for embedded systems, enabling neural network deployment and training on various hardware, including 8-bit microcontrollers and Arduino boards.
Binarized Neural Networks on Microcontrollers
Explore Binarized Neural Networks for efficient deep learning on microcontrollers, enabling complex tasks with minimal memory and high performance using 1-bit encoding for weights and activations.
Security of Edge AI Against Hardware Attacks
Explore security challenges and countermeasures for Edge AI against hardware attacks, focusing on neural network protection and side-channel analysis in embedded systems.
Acceleration of Deep Learning Inference on Raspberry Pi's VideoCore GPU
Explore techniques for accelerating deep learning inference on Raspberry Pi's VideoCore GPU, including optimized tools, libraries, and compilers for enhanced performance without compromising accuracy.
Enabling Neural Network at the Low Power Edge - A Neural Network Compiler
Explore neural network deployment in IoT devices, addressing challenges and optimizing for ultra-low power operation through hardware-software co-optimization and automatic code generation.
Structured Quantization for Neural Network Language Model Compression
探讨神经网络语言模型的结构化量化压缩技术,展示如何在不损失性能的情况下实现70-100倍的高压缩率,适用于资源受限场景。
Powering Innovation in AI Devices on the Edge with Micro-NPUs - tinyML Asia 2020
Explore Arm's Ethos-U micro-NPUs for efficient AI processing on edge devices, enabling new applications and capabilities in embedded systems through advanced Machine Learning solutions.
Networks Within Networks - Novel CNN Design Space Exploration for Resource-Limited Devices
Explore novel CNN design for resource-limited devices using "network within network" paradigm. Learn automated method for compressing neural networks while preserving accuracy and meeting size constraints.
A Technique for Extreme Compression of LSTM Models
Explore extreme LSTM compression using sparse structured additive matrices. Learn doping techniques, co-matrix adaptation, and regularization for state-of-the-art accuracy at high compression factors in NLP applications.
Speech Recognition on Low Power Devices
Explore multilingual speech recognition on low-power devices, covering model optimization techniques, neural network compression, and practical demonstrations for efficient implementation.
Towards Software-Defined Imaging - Adaptive Video Subsampling for Energy-Efficient Object Tracking
Explore adaptive video subsampling for energy-efficient object tracking using software-defined imaging algorithms and CMOS sensors' ROI capabilities, with FPGA-accelerated implementations.
Pushing the Limits of RNN Compression Using Kronecker Products
Explore Kronecker Products for compressing sequence-based neural networks, achieving high compression factors for IoT RNN applications and addressing accuracy loss in large NLP tasks through doped Kronecker products.
Training Neural Networks for Sensors
Overview of training neural networks for gesture recognition using accelerometer data on a low-power device. Covers data collection, augmentation, network architecture, and performance for smartwatch applications.
Low Power CV Meets the Real World
Explore practical considerations for implementing computer vision in low-power IoT devices, covering hardware accelerators, algorithms, optics, and real-world challenges.