TY - GEN
T1 - LoRa-Powered Energy-Effcient Object Detection Mechanism in Edge Computing Nodes
AU - Jindal, Anshul
AU - Jose, Jiby Mariya
AU - Benedict, Shajulin
AU - Gerndt, Michael
N1 - Publisher Copyright:
© 2022 IEEE.
PY - 2022
Y1 - 2022
N2 - The ongoing accomplishments in the decades-long realization of computer vision have infused new dimensions in various research areas such as smart mobility, smart healthcare, education, finance, and so forth. Research works relating to automated object detection, deep learning-assisted data pipelines, and energy-efficient end-to-end solutions have enabled newer perceptions among researchers, albeit the existence of challenges. This paper proposes an object detection system using energy-efficient Long Range (LoRA) communication media on edge nodes such as Raspberry Pi, Coral DevBoard, and Nvidia Jetson Nano. The proposed approach utilized energy-efficient methods to collaboratively offload object detection-related tasks such as capturing images, training images, and inferring objects across a compendium of computing nodes using LoRA. In addition, this research study has attempted to reveal the inference capabilities of images on three different edge nodes. The proposed work has achieved a power difference of at least 1.2 watts during the inference period of the deep learning models without challenging the prediction accuracy with respect to the base model.
AB - The ongoing accomplishments in the decades-long realization of computer vision have infused new dimensions in various research areas such as smart mobility, smart healthcare, education, finance, and so forth. Research works relating to automated object detection, deep learning-assisted data pipelines, and energy-efficient end-to-end solutions have enabled newer perceptions among researchers, albeit the existence of challenges. This paper proposes an object detection system using energy-efficient Long Range (LoRA) communication media on edge nodes such as Raspberry Pi, Coral DevBoard, and Nvidia Jetson Nano. The proposed approach utilized energy-efficient methods to collaboratively offload object detection-related tasks such as capturing images, training images, and inferring objects across a compendium of computing nodes using LoRA. In addition, this research study has attempted to reveal the inference capabilities of images on three different edge nodes. The proposed work has achieved a power difference of at least 1.2 watts during the inference period of the deep learning models without challenging the prediction accuracy with respect to the base model.
KW - Edge
KW - Energy-Efficient
KW - IoT
KW - LoRA
KW - Object Detection
UR - http://www.scopus.com/inward/record.url?scp=85146435472&partnerID=8YFLogxK
U2 - 10.1109/I-SMAC55078.2022.9987393
DO - 10.1109/I-SMAC55078.2022.9987393
M3 - Conference contribution
AN - SCOPUS:85146435472
T3 - 6th International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud), I-SMAC 2022 - Proceedings
SP - 237
EP - 244
BT - 6th International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud), I-SMAC 2022 - Proceedings
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 6th International Conference on I-SMAC (IoT in Social, Mobile, Analytics and Cloud), I-SMAC 2022
Y2 - 10 November 2022 through 12 November 2022
ER -