Naslov
Energy consumption and data amount reduction using object detection on embedded platform

Autori
Šnajder, Boris ; Kalafatić, Zoran ; Bilas, Vedran

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Izvornik
S-CUBE 2016 - 7th EAI International Conference on Sensor Systems and Software / - , 2016, 1-12

Skup
S-CUBE 2016 - 7th EAI International Conference on Sensor Systems and Software

Mjesto i datum
Nica, Francuska, 01.12.2016. - 02.12.2016

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
wireless sensor networks; energy efficiency; image processing; object detection; data amount reduction

Sažetak
High resolution image handling often results with high energy burden for battery-powered devices, such as sensor nodes in WSN. Motivation for this study is assessment of energy consumption of the sensor node with high-resolution camera, featuring image processing. We present a selection of object detection algorithms and evaluate their efficiency. To verify applicability of those algorithms, we acquired image sequence that correspond to applications of pests detection in agriculture. We verified considered algorithms' performances: recall, precision and expected reduction of the data amount. Energy required to execute considered algorithms was measured on ARM processor based platform. Our results show that object extraction on a node can provide reduction of the data amount by up to three orders of magnitude. While simple algorithms can lead to lower overall energy consumption of the node, the more complex algorithm provides better performances, but at a cost of prohibitively high energy consumption.

Izvorni jezik
Engleski

Znanstvena područja
Elektrotehnika, Računarstvo

POVEZANOST RADA