pith. sign in

arxiv: 1906.11056 · v1 · pith:PVRIFGFLnew · submitted 2019-06-26 · 💻 cs.DC

EdgeLens: Deep Learning based Object Detection in Integrated IoT, Fog and Cloud Computing Environments

Shreshth Tuli , Nipam Basumatary , Rajkumar Buyya This is my paper

Pith reviewed 2026-05-25 15:21 UTC · model grok-4.3

classification 💻 cs.DC
keywords object detectionfog computingcloud computingdeep learningIoTlatencyaccuracyEdgeLens
0
0 comments X

The pith

EdgeLens is a framework that lets deep learning object detection switch between high-accuracy and low-latency modes across fog and cloud resources.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper introduces EdgeLens as a way to run deep learning object detection in combined IoT, fog, and cloud settings. It shows the framework can adjust its behavior to favor either higher detection accuracy or shorter response times depending on what the application needs. This addresses the gap where cloud-only solutions add too much delay for tasks like surveillance, and fog resources alone struggle with complex models. Tests measure how the system performs on accuracy, response time, jitter, bandwidth use, and power draw while switching modes.

Core claim

EdgeLens is a framework to deploy deep learning-based object detection in fog-cloud environments that adapts to application or user requirements to provide either high accuracy or low latency modes of service.

What carries the argument

The EdgeLens framework, which allocates deep learning inference tasks across edge, fog, and cloud layers and switches operating modes to meet different service quality goals.

If this is right

  • Object detection services can run in integrated IoT-fog-cloud setups without needing separate custom builds for each performance target.
  • Response time, jitter, bandwidth consumption, and power use become adjustable by choosing the accuracy or latency mode.
  • Time-sensitive IoT applications gain access to deep learning capabilities that pure cloud deployments could not deliver due to network delay.
  • The same framework can support multiple service levels for different users or applications on shared infrastructure.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The adaptation approach could extend to other deep learning workloads such as video analytics or anomaly detection that also face latency-accuracy trade-offs in distributed systems.
  • Widespread use might reduce reliance on specialized edge hardware by letting existing cloud and fog nodes handle parts of the computation on demand.
  • Scalability questions arise around handling many simultaneous streams or dynamic network changes, which could be tested in larger simulated environments.

Load-bearing premise

Deep learning models for object detection can be successfully partitioned and run across distributed fog and cloud resources while keeping accuracy usable and response times short enough for real applications.

What would settle it

A deployment of EdgeLens on real IoT cameras for surveillance where the high-accuracy mode drops below 80 percent correct detections or the low-latency mode exceeds 200 milliseconds average response time under typical network conditions.

Figures

Figures reproduced from arXiv: 1906.11056 by Nipam Basumatary, Rajkumar Buyya, Shreshth Tuli.

Figure 1
Figure 1. Figure 1: Architecture of proposed system [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 3
Figure 3. Figure 3: Android Interface at gateway device [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Experiment setup [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: mAP results [PITH_FULL_IMAGE:figures/full_fig_p005_5.png] view at source ↗
read the original abstract

Data-intensive applications are growing at an increasing rate and there is a growing need to solve scalability and high-performance issues in them. By the advent of Cloud computing paradigm, it became possible to harness remote resources to build and deploy these applications. In recent years, new set of applications and services based on Internet of Things (IoT) paradigm, require to process large amount of data in very less time. Among them surveillance and object detection have gained prime importance, but cloud is unable to bring down the network latencies to meet the response time requirements. This problem is solved by Fog computing which harnesses resources in the edge of the network along with remote cloud resources as required. However, there is still a lack of frameworks that are successfully able to integrate sophisticated software and applications, especially deep learning, with fog and cloud computing environments. In this work, we propose a framework to deploy deep learning-based applications in fog-cloud environments to harness edge and cloud resources to provide better service quality for such applications. Our proposed framework, called EdgeLens, adapts to the application or user requirements to provide high accuracy or low latency modes of services. We also tested the performance of the software in terms of accuracy, response time, jitter, network bandwidth and power consumption and show how EdgeLens adapts to different service requirements.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 0 minor

Summary. The manuscript proposes EdgeLens, a framework for deploying deep learning-based object detection in integrated IoT, fog, and cloud environments. It claims the framework adapts to application or user requirements to switch between high-accuracy and low-latency service modes, and reports performance tests on accuracy, response time, jitter, bandwidth, and power consumption.

Significance. If the adaptation mechanism is implemented and the reported performance results hold under rigorous testing, the work could provide a practical contribution to edge-cloud integration for latency-sensitive DL applications such as surveillance. The absence of any equations, fitted parameters, or reproducible experimental details, however, prevents assessment of whether the claimed adaptation is novel or merely a high-level orchestration layer.

major comments (1)
  1. [Abstract] Abstract: The abstract states that performance tests were conducted on accuracy, response time, jitter, bandwidth and power, but provides no methods, data, error bars or exclusion criteria; the central adaptation claim cannot be verified from the given text.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their review. We address the comment on the abstract below, noting that abstracts are concise by design while full details appear in the manuscript body.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The abstract states that performance tests were conducted on accuracy, response time, jitter, bandwidth and power, but provides no methods, data, error bars or exclusion criteria; the central adaptation claim cannot be verified from the given text.

    Authors: Abstracts are by nature concise summaries and do not typically include detailed methods, data, error bars, or exclusion criteria, which are instead provided in the body of the paper. The EdgeLens framework and its adaptation mechanism for switching between high-accuracy and low-latency modes are described in detail in Section 3 of the manuscript. The performance evaluation, including the test methodology, results for accuracy, response time, jitter, bandwidth, and power consumption, along with any relevant data and figures, are presented in Sections 4 and 5. This allows readers to verify the claims from the full text. revision: no

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper proposes the EdgeLens framework for deploying deep learning object detection in integrated IoT/fog/cloud settings and reports empirical measurements of accuracy, latency, jitter, bandwidth and power under high-accuracy versus low-latency modes. No equations, fitted parameters, predictions derived from prior fits, or load-bearing self-citations appear in the provided text. The central claims rest on the existence and measured behavior of an implemented adaptation mechanism rather than any derivation that reduces to its own inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

No free parameters, axioms, or invented entities are identifiable from the abstract alone.

pith-pipeline@v0.9.0 · 5771 in / 1044 out tokens · 27143 ms · 2026-05-25T15:21:02.512523+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

20 extracted references · 20 canonical work pages · 3 internal anchors

  1. [1]

    and Kwak K.S.: The internet of things for health care: a comprehensive survey

    Islam, S.M.R., Kwak, D., Kabir M.D.H., Hossain M. and Kwak K.S.: The internet of things for health care: a comprehensive survey. IEEE Access . 3, 678-708 (2015)

    work page 2015

  2. [2]

    Real time detection of speed breakers and warning system for on-road drivers

    Afrin, Mahbuba, Md Redowan Mahmud, and Md Abdur Razzaque. "Real time detection of speed breakers and warning system for on-road drivers." 2015 IEEE International WIE Conference on Electrical and Computer Engineering (WIECON-ECE), pp. 495-498. IEEE, 2015

    work page 2015

  3. [3]

    FogBus: A Blockchain-based Lightweight Framework for Edge and Fog Computing

    Tuli, Shreshth, Redowan Mahmud, Shikhar Tuli, and Rajkumar Buyya. "FogBus: A Blockchain-based Lightweight Framework for Edge and Fog Computing." Journal of Systems and Software (2019)

    work page 2019

  4. [4]

    Fog computing and its role in the internet of things

    Bonomi, Flavio, Rodolfo Milito, Jiang Zhu, and Sateesh Addepalli. "Fog computing and its role in the internet of things." MCC workshop on Mobile cloud computing, pp. 13-16. ACM, 2012

    work page 2012

  5. [5]

    Salient Object Detection: A Survey

    Borji, Ali, Ming -Ming Cheng, Qibin Hou, Huaizu Jiang, and Jia Li. "Salient object detection: A survey." arXiv:1411.5878 (2014)

    work page internal anchor Pith review Pith/arXiv arXiv 2014

  6. [6]

    Deep learning: the frontier for distributed attack detection in fog -to-things computing

    Abeshu, Abebe, and Naveen Chilamkurti. "Deep learning: the frontier for distributed attack detection in fog -to-things computing." IEEE Communications Magazine 56, no. 2 (2018): 169-175

    work page 2018

  7. [7]

    Aneka: a software platform for .NET -based cloud computing

    Vecchiola, Christian, Xingchen Chu, and Rajkumar Buyya. "Aneka: a software platform for .NET -based cloud computing." High Speed and Large Scale Scientific Computing 18 (2009): 267-295

    work page 2009

  8. [8]

    Smart urban surveillance using fog computing

    Chen, Ning, Yu Chen, Sejun Song, Chin -Tser Huang, and Xinyue Ye. "Smart urban surveillance using fog computing." 2016 IEEE/ACM Symposium on Edge Computing (SEC), pp. 95-96. IEEE, 2016

    work page 2016

  9. [9]

    Distributed attack detection scheme using deep learning approach for Internet of Things

    Diro, Abebe Abeshu, and Naveen Chilamkurti. "Distributed attack detection scheme using deep learning approach for Internet of Things." Future Generation Computer Systems 82 (2018): 761-768

    work page 2018

  10. [10]

    Deep learning for smart industry: efficient manufacture inspection system with fog com puting

    Li, Liangzhi, Kaoru Ota, and Mianxiong Dong. "Deep learning for smart industry: efficient manufacture inspection system with fog com puting." IEEE Transactions on Industrial Informatics (2018): 4665-4673

    work page 2018

  11. [11]

    Distributed deep neural networks over the cloud, the edge and end devices

    Teerapittayanon, Surat, Bradley McDanel, and Hsiang -Tsung Kung. "Distributed deep neural networks over the cloud, the edge and end devices." 37 th IEEE International Conference on Distributed Computing Systems (ICDCS), pp. 328-339. IEEE, 2017

    work page 2017

  12. [12]

    Fog-Assisted wIoT: A Smart Fog Gateway for End-to-End Analytics in Wearable Internet of Things

    Constant, Nic holas, Debanjan Borthakur, Mohammadreza Abtahi, Harishchandra Dubey, and Kunal Mankodiya. "Fog -assisted wiot: A smart fog gateway for end-to-end analytics in wearable internet of things." arXiv preprint arXiv:1701.08680 (2017)

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  13. [13]

    http://appinventor.mit.edu/appinventor- sources/

    MIT App Inventor software. http://appinventor.mit.edu/appinventor- sources/. [accessed on 28-May-2019]

    work page 2019

  14. [14]

    You only look once: Unified, real-time object detection

    Redmon, Joseph, Santosh Divvala, Ross Girshick, and Ali Farhadi. "You only look once: Unified, real-time object detection." IEEE conference on Computer Vision and Pattern Recognition, pp. 779-788. 2016

    work page 2016

  15. [15]

    YOLOv3: An Incremental Improvement

    Redmon, Joseph, and Ali Farhadi. "Yolov3: An incremental improvement." arXiv preprint arXiv:1804.02767 (2018)

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  16. [16]

    http://cocodataset.org/#home

    COCO Dataset. http://cocodataset.org/#home . [accessed on 31-May- 2019]

    work page 2019

  17. [17]

    https://docs.microsoft.com/ enus/windows-hardware/test/wpt/

    Mircosoft Windows performance toolkit . https://docs.microsoft.com/ enus/windows-hardware/test/wpt/. [accessed on 30-May-2019]

    work page 2019

  18. [18]

    https://www.microsoft.com/enau/download/details.aspx?id=4865

    Microsoft Network Monitor 3.4. https://www.microsoft.com/enau/download/details.aspx?id=4865. [accessed on 28-May-2019]

    work page 2019

  19. [19]

    Advanced deep -learning techniques for salient and category -specific object detection: a survey

    Han, Junwei, Dingwen Zhang, Gong Cheng, Nian Liu, and Dong Xu. "Advanced deep -learning techniques for salient and category -specific object detection: a survey." IEEE Signal Processing Magazine 35, no. 1 (2018): 84-100

    work page 2018

  20. [20]

    The pascal visual object classes (voc) challenge

    Everingham, Mark, Luc Van Gool, Christopher KI Williams, John Winn, and Andrew Zisserman. "The pascal visual object classes (voc) challenge." International journal of computer vision 88, no. 2 (2010): 303-338

    work page 2010