Eyes of Things

Abstract

Embedded systems control and monitor a great deal of our reality. While some “classic” features are intrinsically necessary, such as low power consumption, rugged operating ranges, fast response and low cost, these systems have evolved in the last few years to emphasize connectivity functions, thus contributing to the Internet of Things paradigm. A myriad of sensing/computing devices are being attached to everyday objects, each able to send and receive data and to act as a unique node in the Internet. Apart from the obvious necessity to process at least some data at the edge (to increase security and reduce power consumption and latency), a major breakthrough will arguably come when such devices are endowed with some level of autonomous “intelligence”. Intelligent computing aims to solve problems for which no efficient exact algorithm can exist or for which we cannot conceive an exact algorithm. Central to such intelligence is Computer Vision (CV), i.e., extracting meaning from images and video. While not everything needs CV, visual information is the richest source of information about the real world: people, places and things. The possibilities of embedded CV are endless if we consider new applications and technologies, such as deep learning, drones, home robotics, intelligent surveillance, intelligent toys, wearable cameras, etc. This paper describes the Eyes of Things (EoT) platform, a versatile computer vision platform tackling those challenges and opportunities.

Automatic Museum Audio Guide

“…Finally, USB and micro-SD encrypted functions allow fast and secure data logging. A detailed description of the EoT board can be found in [23].…”

“…In the context described above, the eyes of things (EoT) project designed a general-purpose small embedded device for computer vision that includes the Intel Movidius Myriad2 processor. The entire project was based on four fundamental pillars: Cost, power consumption, size, and programming flexibility [22,23]. This platform is optimized to maximize inferred information per milliwatt and adapt the quality of inferred results to each particular application.…”

Efficient Winograd-based Convolution Kernel Implementation on Edge Devices

“…The equation Z = A T YA that produces the 2x2 convolution output (step (5) in the description of the Winograd algorithm) can be written as follows (details can be found in [9]):…”

“…A recent trend towards the deployment of neural networks in edge devices is the design of CNNs with limited requirements in computational resources, such as the SqueezeNet [8] [5]. Another approach, which is complementary to this and in which this work focuses, is to develop techniques and methodologies for the efficient implementation of CNNs in such systems.…”

Smart Doll: Emotion Recognition Using Embedded Deep Learning

“…The Eyes of Things (EoT) [4] platform is an optimized core vision unit designed by the authors, that can work independently and also embedded into all types of artefacts ( Figure 1) at minimal power consumption. The EoT platform is based on the Myriad 2 chip by Movidius [5,6].…”