PETRI: Reducing Bandwidth Requirement in Smart Surveillance by Edge-Cloud Collaborative Adaptive Frame Clustering and Pipelined Bidirectional Tracking

Neural networks running on cloud servers have been widely used in smart surveillance, but they require high bandwidth to upload videos. Edge-cloud collaborative encoding based on ROI (Region-Of-Interest) can reduce bandwidth requirement, but it suffers from inaccurate ROI detection due to feedback l...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:2021 58th ACM/IEEE Design Automation Conference (DAC) s. 421 - 426
Hlavní autoři: Liu, Ruoyang, Zhang, Lu, Wang, Jingyu, Yang, Huazhong, Liu, Yongpan
Médium: Konferenční příspěvek
Jazyk:angličtina
Vydáno: IEEE 05.12.2021
Témata:
Bandwidth
Collaboration
collaborative architecture
edge computing
Image edge detection
Object detection
Pipelines
Surveillance
Target tracking
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:Neural networks running on cloud servers have been widely used in smart surveillance, but they require high bandwidth to upload videos. Edge-cloud collaborative encoding based on ROI (Region-Of-Interest) can reduce bandwidth requirement, but it suffers from inaccurate ROI detection due to feedback latency and undetected new targets. To address the above challenges, we propose an object detection system named PETRI. It adopts a latency-hiding pipeline workflow with adaptive keyframe interval selection for different input videos, and utilizes a retro-tracking method to find undetected targets. While achieving negligible impact on model accuracy, the proposed PETRI can save up to 66.44% and 30.25% bandwidth compared with the cloud only method and the previous state-of-art work respectively.
DOI:10.1109/DAC18074.2021.9586088