Accelerated Visual Context Classification on a Low-Power Smartwatch
- 18 November 2016
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Human-Machine Systems
- Vol. 47 (21682291) , 1-12
Abstract
Data produced by wearable sensors is key in contexts such as performance enhancement and training help for sports and fitness, continuous monitoring for aging people and for chronic disease management, and in gaming and entertainment. Unfortunately, wearable devices currently in the market are either incapable of complex functionality or severely impaired by short battery lifetime. In this work, we present a smart watch platform based on an ultralow-power (ULP) heterogeneous system composed of a TI MSP430 microcontroller, the PULP programmable parallel accelerator, and a set of ULP sensors, including a camera. The embedded PULP accelerator enables state-of-the-art context classification based on convolutional neural networks to be applied within a sub-10-mW system power envelope. Our methodology enables to reach high accuracy in context classification over five classes (up to 84%, with three classes over five reaching more than 90% accuracy), while consuming 2.2 mJ per classification, or an ultralow energy consumption of less than 91 μJ per classification with an accuracy of 64%-3.2× better than chance. Our results suggest that the proposed heterogeneous platform can provide up to 500× speedup with respect to the MSP430 within a similar power envelope, which would enable complex computer vision algorithms to be executed in highly power-constrained scenarios.Keywords
Funding Information
- MicroLearn: Micropower Deep Learning Swiss SNF
- IcySoC and YINS RTD
- Swiss SNF
- Nano-Tera.ch with Swiss Confederation financing
- Horizon 2020 ExaNoDe Project (671578)
This publication has 53 references indexed in Scilit:
- Hexagon DSP: An Architecture Optimized for Mobile Multimedia and CommunicationsIEEE Micro, 2014
- Internet of Things (IoT): A vision, architectural elements, and future directionsFuture Generation Computer Systems, 2013
- Ultra low-power signal processing in wearable monitoring systemsACM Transactions on Embedded Computing Systems, 2013
- SleepWalker: A 25-MHz 0.4-V Sub-$\hbox{mm}^{2}$ 7-$\mu\hbox{W/MHz}$ Microcontroller in 65-nm LP/GP CMOS for Low-Carbon Wireless Sensor NodesIEEE Journal of Solid-State Circuits, 2012
- Face recognition using Histograms of Oriented GradientsPattern Recognition Letters, 2011
- SurroundSenseACM SIGMOBILE Mobile Computing and Communications Review, 2009
- Cross-Validation2009
- Top 10 algorithms in data miningKnowledge and Information Systems, 2007
- A low-power wireless video sensor node for distributed object detectionJournal of Real-Time Image Processing, 2007
- Gradient-based learning applied to document recognitionProceedings of the IEEE, 1998