A Software/Hardware Co-Design Framework for the ‘Internet of Eyes’

This paper examines the challenges involved in bringing real-time image analysis to the Internet of Things (IoT) and thereby develops a software/hardware co-design framework that takes account of the power and computational requirements of IoT edge devices. Current state-of-art solutions are typically suited to applications that require low power or low latency image analysis, but not both. This paper describes an architecture that can be used to perform low latency, low power processing on an edge device, thus making it suitable for wireless ‘Internet of Eyes’ (IoE) applications. Such applications require the processing of vast amounts of high-resolution video data in order to extract small amounts of salient information, which can then be transmitted to cloud platforms using low-bandwidth network communications protocols. This novel approach is tested by applying it to a real-time vision-based distributed motorway vehicle counting application, and evaluated for suitability to an energy harvesting deployment.

[1]  Reinhard Klette,et al.  Traffic intersection monitoring using fusion of GMM-based deep learning classification and geometric warping , 2017, 2017 International Conference on Image and Vision Computing New Zealand (IVCNZ).

[2]  Nattha Jindapetch,et al.  SDSoC based development of vehicle counting system using adaptive background method , 2017, 2017 IEEE Regional Symposium on Micro and Nanoelectronics (RSM).

[3]  Steve B. Furber,et al.  Power analysis of large-scale, real-time neural networks on SpiNNaker , 2013, The 2013 International Joint Conference on Neural Networks (IJCNN).

[4]  Muhammad Atif,et al.  Evaluation of High Density GPUs as Sustainable Smart City Infrastructure , 2015, 2015 IEEE/ACM 8th International Conference on Utility and Cloud Computing (UCC).

[5]  Geoffrey C. Fox,et al.  A Framework for Real Time Processing of Sensor Data in the Cloud , 2015, J. Sensors.

[6]  Grzegorz Bieszczad.  SoC-FPGA embedded system for real-time thermal image processing , 2016, 2016 MIXDES - 23rd International Conference Mixed Design of Integrated Circuits and Systems.