• Stars
    star
    1,040
  • Rank 42,621 (Top 0.9 %)
  • Language
    C
  • License
    Apache License 2.0
  • Created over 6 years ago
  • Updated 11 months ago

Reviews

There are no reviews yet. Be the first to send feedback to the community and the maintainers!

Repository Details

Keyword spotting on Arm Cortex-M Microcontrollers

Keyword spotting for Microcontrollers

This repository consists of the tensorflow models and training scripts used in the paper: Hello Edge: Keyword spotting on Microcontrollers. The scripts are adapted from Tensorflow examples and some are repeated here for the sake of making these scripts self-contained.

To train a DNN with 3 fully-connected layers with 128 neurons in each layer, run:

python train.py --model_architecture dnn --model_size_info 128 128 128 

The command line argument --model_size_info is used to pass the neural network layer dimensions such as number of layers, convolution filter size/stride as a list to models.py, which builds the tensorflow graph based on the provided model architecture and layer dimensions. For more info on model_size_info for each network architecture see models.py. The training commands with all the hyperparameters to reproduce the models shown in the paper are given here.

To run inference on the trained model from a checkpoint on train/val/test set, run:

python test.py --model_architecture dnn --model_size_info 128 128 128 --checkpoint 
<checkpoint path>

To freeze the trained model checkpoint into a .pb file, run:

python freeze.py --model_architecture dnn --model_size_info 128 128 128 --checkpoint 
<checkpoint path> --output_file dnn.pb

Pretrained models

Trained models (.pb files) for different neural network architectures such as DNN, CNN, Basic LSTM, LSTM, GRU, CRNN and DS-CNN shown in this arXiv paper are added in Pretrained_models. Accuracy of the models on validation set, their memory requirements and operations per inference are also summarized in the following table.

To run an audio file through the trained model (e.g. a DNN) and get top prediction, run:

python label_wav.py --wav <audio file> --graph Pretrained_models/DNN/DNN_S.pb 
--labels Pretrained_models/labels.txt --how_many_labels 1

Quantization Guide and Deployment on Microcontrollers

A quick guide on quantizing the KWS neural network models is here. The example code for running a DNN model on a Cortex-M development board is also provided here.

More Repositories

1

ComputeLibrary

The Compute Library is a set of computer vision and machine learning functions optimised for both Arm CPUs and GPUs using SIMD technologies.
C++
2,539
star
2

arm-trusted-firmware

Read-only mirror of Trusted Firmware-A
C
1,690
star
3

CMSIS_5

CMSIS Version 5 Development Repository
C
1,233
star
4

armnn

Arm NN ML Software. The code here is a read-only mirror of https://review.mlplatform.org/admin/repos/ml/armnn
C++
1,104
star
5

astc-encoder

The Arm ASTC Encoder, a compressor for the Adaptive Scalable Texture Compression data format.
C
880
star
6

abi-aa

Application Binary Interface for the Arm® Architecture
HTML
673
star
7

vulkan_best_practice_for_mobile_developers

Vulkan best practice for mobile developers
C++
564
star
8

optimized-routines

Optimized implementations of various library functions for ARM architecture processors
C
486
star
9

CMSIS-FreeRTOS

FreeRTOS adaptation for CMSIS-RTOS Version 2
C
460
star
10

CMSIS_4

Cortex Microcontroller Software Interface Standard (V4 no longer maintained)
C
437
star
11

ML-examples

Arm Machine Learning tutorials and examples
C++
371
star
12

LLVM-embedded-toolchain-for-Arm

A project dedicated to building LLVM toolchain for 32-bit Arm embedded targets.
CMake
331
star
13

opengl-es-sdk-for-android

OpenGL ES SDK for Android
CSS
325
star
14

mango

Parallel Hyperparameter Tuning in Python
Jupyter Notebook
310
star
15

SCALE-Sim

Python
296
star
16

CMSIS-DSP

CMSIS-DSP embedded compute library for Cortex-M and Cortex-A
C
277
star
17

Arm-2D

2D Graphic Library optimized for Cortex-M processors
C
232
star
18

Tool-Solutions

Tutorials & examples for Arm software development tools.
C
217
star
19

EndpointAI

C++
216
star
20

SCP-firmware

Read-only mirror of System Control Processor (SCP) firmware
C
205
star
21

vulkan-sdk

Github repository for the Vulkan SDK
C
199
star
22

lisa

Linux Integrated System Analysis
Jupyter Notebook
192
star
23

HWCPipe

Hardware counters interface
C++
188
star
24

u-boot

Clone of upstream U-Boot repo with patches for Arm development boards
C
169
star
25

ML-zoo

Python
149
star
26

android-nn-driver

C++
149
star
27

CMSIS-NN

CMSIS-NN Library
C
148
star
28

CMSIS-Driver

Repository of microcontroller peripheral driver implementing the CMSIS-Driver API specification
C
144
star
29

workload-automation

A framework for automating workload execution and measurement collection on ARM devices.
Python
138
star
30

gator

Sources for Arm Streamline's gator daemon
C++
121
star
31

perfdoc

A cross-platform Vulkan layer which checks Vulkan applications for best practices on Arm Mali devices.
C++
112
star
32

ebbr

Embedded Base Boot Requirements Specification
PostScript
108
star
33

keyword-transformer

Official implementation of the Keyword Transformer: https://arxiv.org/abs/2104.00769
Jupyter Notebook
108
star
34

linux

C
95
star
35

asl-interpreter

Example implementation of Arm's Architecture Specification Language (ASL)
OCaml
94
star
36

mobile-studio-integration-for-unity

Mobile Studio tool integration with C# scripting for the Unity game engine.
C
86
star
37

sbsa-acs

ARM Enterprise: SBSA Architecture Compliance Suite
C
84
star
38

CMSIS_6

CMSIS version 6 (successor of CMSIS_5)
C
83
star
39

sesr

Super-Efficient Super Resolution
Python
80
star
40

CSAL

Coresight Access Library
C
78
star
41

progress64

PROGRESS64 is a C library of scalable functions for concurrent programs, primarily focused on networking applications.
C
65
star
42

Cloud-IoT-Core-Kit-Examples

Example projects and code are supplied to support the Arm-based IoT Kit for Cloud IoT Core
Python
61
star
43

trappy

This repository has moved to https://gitlab.arm.com/tooling/trappy
Python
61
star
44

psa-arch-tests

Tests for verifying implementations of TBSA-v8M and the PSA Certified APIs
C
61
star
45

cmsis-pack-eclipse

CMSIS-Pack Eclipse Plug-ins
Java
60
star
46

developer

GTM related documentation
C++
59
star
47

AVH-GetStarted

C
56
star
48

ethos-n-driver-stack

Driver stack (including user space libraries, kernel module and firmware) for the Arm® Ethos™-N NPU
C++
53
star
49

acle

Arm C Language Extensions (ACLE)
Python
52
star
50

patrace

C++
52
star
51

tarmac-trace-utilities

Tools for analyzing and browsing Tarmac instruction traces.
C++
47
star
52

devlib

Library for interaction with and instrumentation of remote devices.
Python
44
star
53

speculation-barrier

This project provides a header file which contains wrapper macros for the __builtin_load_no_speculate builtin function defined at https://www.arm.com/security-update This builtin function defines a speculation barrier, which can be used to limit the conditions under which a value which has been loaded can be used under speculative execution.
Objective-C
44
star
54

arm-systemready

Arm SystemReady
Shell
43
star
55

arm-enterprise-acs

ARM Enterprise ACS
C
39
star
56

DeepFreeze

SystemVerilog
38
star
57

tf-issues

Issue tracking for the ARM Trusted Firmware project
36
star
58

scalpel

This is a PyTorch implementation of the Scalpel. Node pruning for five benchmark networks and SIMD-aware weight pruning for LeNet-300-100 and LeNet-5 is included.
Python
35
star
59

psa-api

Documentation source and development of the PSA Certified API
C
34
star
60

perf-libs-tools

C
31
star
61

bob-build

Meta-build system using Blueprint and ninja
Go
30
star
62

CMSIS-RTX

RTX5 real time kernel for Arm Cortex-based embedded systems (spin-off from CMSIS_5)
C
29
star
63

TZ-TRNG

TrustZone True Number Generator
C
29
star
64

mram_simulation_framework

MRAM magnetization simulation framework. s-LLGS python and verilog-a solvers for transients simulation and Fokker-planck equation solver for stochastic analysis
Python
28
star
65

AVH

AVH FVPs: Arm Virtual Hardware with Fixed Virtual Platforms
C
27
star
66

bento-linker

A light-weight alternative to processes for microcontrollers.
C
27
star
67

data

Machine-readable data describing Arm architecture and implementations. Includes JSON descriptions of implemented PMU events.
26
star
68

synchronization-benchmarks

Collection of synchronization micro-benchmarks and traces from infrastructure applications
C
26
star
69

libGPUInfo

A utility library for application developers to query the configuration of the Arm Immortalis GPU or Arm Mali GPU present in their system.
C++
24
star
70

toolchain-gnu-bare-metal

A toolchain sub-project dedicated to build GNU toolchain for 32-bit bare-metal targets
Shell
24
star
71

NXP_LPC

CMSIS Driver Implementations for the NXP LPC Microcontroller Series
C
23
star
72

golang-utils

Helpers and utilities for Golang in order to do actions not available in the standard library.
Go
23
star
73

libddssec

DDS Security library - Project moved to https://gitlab.arm.com/libraries/libddssec
C
23
star
74

AArch64cryptolib

AArch64cryptolib is a from scratch implementation of cryptographic primitives aiming for optimal performance on Arm A-class cores
C
23
star
75

cryptocell-312-runtime

CryptoCell 312 runtime code
C
22
star
76

Shackleton-Framework

A generic genetic programming framework that aims to make genetic programming easier for a myriad of uses. Currently, the main target is to use the framework for code optimization in tandem with the LLVM framework.
C
22
star
77

CMSIS-View

Repository of CMSIS Software Pack for software event generation and input/output handling.
Go
22
star
78

AVH-TFLmicrospeech

Example: Micro speech for TensorFlow Lite
C
21
star
79

PAF

PAF (the Physical Attack Framework) is a framework for analyzing physical attacks: fault injection and side channels
C++
20
star
80

bart

Behavioural Analysis and Regression Toolkit
Python
20
star
81

HPCG_for_Arm

C++
20
star
82

armnn-mlperf

Arm mlperf.org benchmark port
C++
20
star
83

coresight-wire-protocol

Coresight Wire Protocol (CSWP) Server/Client and streaming trace examples.
HTML
18
star
84

ATP-Engine

C++
18
star
85

ATS-Keyword

Smart Home Total Solution - Keyword Recognition
C
17
star
86

vscode-cmsis-csolution

Extension support for VS Code CMSIS Project Extension
17
star
87

vscode-keil-studio-pack

Extension pack for all VS Code extensions
16
star
88

vr-sdk-for-android

VR SDK for Android
CSS
16
star
89

meabo

Multi-purpose multi-phase micro-benchmark
C
15
star
90

avhclient

Arm Virtual Hardware Client
Python
15
star
91

CMSIS-Compiler

CMSIS Compiler support for Arm Compiler
C
15
star
92

Methodology_for_ArmIE_SVE

C++
15
star
93

vktrace-arm

Vktrace arm fork
C++
14
star
94

bsa-acs

Arm SystemReady : BSA Architecture Compliance Suite
C
14
star
95

open-iot-sdk

Open-IoT-SDK - Home of the Total Solution applications.
C
14
star
96

CMSIS-RTOS2_Validation

Validation test suite for CMSIS-RTOS2 API implementations using Arm Virtual Hardware (AVH).
C
14
star
97

CMSIS-Driver_Validation

Test suite for verifying CMSIS-Driver implementations.
C
13
star
98

drm-hwcomposer

Clone of the drm-hwcomposer from freedesktop.org plus patches from Mali DP team
C++
12
star
99

nomali-model

A simple Mali 6xx/7xx register interface model that doesn't do any rendering.
C++
11
star
100

Cortex_DFP

CMSIS generic Arm Cortex-M device family pack
C
11
star