Pith. sign in

REVIEW 4 major objections 5 minor 27 references

A Compendium of Autonomous Navigation using Object Detection and Tracking in Unmanned Aerial Vehicles

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper claims that the main limits of radio-controlled UAVs—signal range, real-time processing, the need for a skilled pilot, hardware cost, and data security—can be solved by making drones autonomous through computer-vision object…

desk verdict A shallow, error-prone literature list that makes a load-bearing claim its own cited papers contradict; no new result, no analysis, and not worth referee time. read the letter →

arxiv 2506.05378 v1 pith:LBWHP5EB submitted 2025-05-31 cs.CV cs.ROeess.IV

Mohit Arora , Pratyush Shukla , Shivali Chopra This is my paper
classification cs.CVcs.ROeess.IV
keywords ComputerVisionAutonomousNavigationObjectDetectionTrackingUnmannedAerialVehiclesDeepLearningDroneReal-TimeProcessing
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This is a survey paper, not a report of new experiments. It gathers a set of existing computer-vision approaches for detecting and tracking objects from a UAV's camera and argues that these methods solve the challenges that limit radio-controlled UAVs: signal quality and range, real-time processing, the need for a skilled pilot, hardware constraints, and data security. The central claim is that programming UAVs to be autonomous through object detection and tracking makes them capable of navigating complex environments without GPS or extra sensor boards, at a cost and accuracy suitable for commercial and research use. If the claim holds, drones for disaster management, dense-area exploration, and traffic surveillance could rely on a camera and efficient onboard algorithms rather than on ground controllers and heavy sensing payloads.

What carries the argument

The central mechanism is the object-detection-and-tracking pipeline: a live camera feed from the UAV is processed by a deep-learning detector (SSD/MobileNet, RefineDet, SegNet) or a tracker (kernelized correlation filter, sparse representation, coarse-to-fine reinforcement learning) to produce identities and locations of objects in each frame. This pipeline is what the paper claims replaces GPS, ground controllers, and external sensor modules. It works by converting the video stream into immediate, hardware-suitable judgments—what an object is and where it is—so the flight controller can react in real time, and it is carried by pretrained models and open-source libraries that keep the computation light enough for the drone's onboard processors.

What would settle it

Run a quadcopter in a cluttered, GPS-denied indoor space or forest, steering it only from the output of a real-time object detector and tracker, with no inertial navigation, no waypoints, and no external planning module. If it cannot traverse a safe course or requires an added sensor or planner to avoid collision, the paper's sufficiency claim collapses.

Watch

Extended reading notes

Core claim

The paper reports a survey-level finding: computer-vision object detection and tracking has reached the point where it can serve as the basis for autonomous UAV navigation in real time. Across the reviewed systems, deep-learning detectors such as SSD with MobileNet and RefineDet, sparse-representation and kernelized-correlation-filter trackers, and stereo-vision SLAM detect vehicles, humans, aircraft, and obstacles from drone cameras, with reported accuracies up to 97.3% for human detection in forest and 92% for vehicle detection on a collected dataset. The paper concludes from these results that algorithms can be tailored to the limited computing power a drone can carry, so that a UAV can detect and infer the kind of objects in its path, reducing dependency on external sensors and boards, and making autonomous navigation economical, robust, and accurate.

Load-bearing premise

The load-bearing premise is that detecting and tracking objects in a camera feed is sufficient for a UAV to navigate autonomously in complex environments, without GPS, sensor fusion, or a planning layer.

Editorial extensions

If this is right

  • Disaster-management and search-and-rescue drones could operate without GPS or a ground pilot, using camera detection of people and obstacles to guide flight.
  • Vision-only sense-and-avoid could become standard: the survey's KCF-based obstacle tracker refines the target whenever tracker confidence drops, giving a real-time collision-avoidance mechanism.
  • Small drones with modest onboard computing could run efficient models like MobileNet-SSD and still achieve real-time detection, lowering the hardware cost of autonomy.
  • Urban traffic surveillance could be performed by UAVs: the surveyed vehicle detectors report real-time accuracy of 92% on collected data and 90.4% on a public drone dataset.
  • Long autonomous flights could maintain target lock despite occlusion, camera shake, and changing aspect ratios, because UAV-specific trackers handle those disturbances.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • An inference beyond the paper: the surveyed pieces form a toolkit, not a full autopilot; coupling the vision trackers with a path-planning layer, such as the graph-based A* planner the paper mentions only for GPS-based flight, would produce the end-to-end autonomous stack the introduction promises.
  • An inference beyond the paper: the reported figures measure perception accuracy, not navigation success; a closed-loop metric such as missions completed per battery charge or obstacle-avoidance rate would test whether detection and tracking really translate into autonomous flight.
  • An inference beyond the paper: the motivation's distinction between detecting an object and inferring whether it is an obstacle or threat suggests a testable extension in which the same detector output is mapped to risk classes, enabling the same camera feed to drive both navigation and security decisions without extra hardware.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. This manuscript is a short survey-style paper claiming that the operational challenges of UAVs—signal quality and range, real-time processing, human expertise, robust hardware, and data security—'can be solved by programming UAVs to be autonomous, using object detection and tracking, through Computer Vision algorithms.' The paper motivates this claim in Sections 1 and 2, then summarizes roughly eighteen papers from 2018–2019 in Section 3, with a consolidated table (Table 1) of architectures, datasets, and findings. It concludes that advanced computer-vision algorithms can provide accurate and efficient autonomous navigation subject to hardware limitations. The paper contains no original measurements, no formal survey methodology, and no critical synthesis; its central claim is both unsupported by the cited evidence and internally inconsistent with several surveyed systems that rely on SLAM, planning, or regression rather than detection/tracking alone.

Significance. If the central claim were established, the paper would provide a useful entry point to the UAV navigation literature and a concise justification for vision-based autonomy. The bibliography is genuinely relevant and includes several well-known works on UAV object detection and tracking, and the tabular summary may help readers locate primary sources. However, the survey's value is severely limited by the absence of a defined selection methodology, the lack of verification or error analysis for the copied accuracy numbers, and the unsupported sufficiency claim that detection and tracking alone can solve all listed challenges. The paper offers no reproducible artifacts, no falsifiable predictions, and no comparative evaluation, so its contribution as a research or survey article is currently not established.

major comments (4)
  1. [Abstract and Section 1] The paper's headline assertion is that the listed challenges—signal quality and range, real-time processing, human expertise, robust hardware, and data security—'can be solved by programming UAVs to be autonomous, using object detection and tracking, through Computer Vision algorithms.' The survey never explains how object detection and tracking address signal quality/range or data security, and no cited paper makes such a connection. The conclusion therefore does not follow from the evidence presented.
  2. [Section 3 and Table 1] The surveyed systems actually undermine the sufficiency claim: Valenti et al. [20] rely on stereo SLAM, local obstacle grids, and an Extended Kalman Filter; Maciel-Pearson et al. [11] use end-to-end regression without explicit object detection; and Table 1 itself describes Kwak et al. [23] as using graph-based A* search 'instead of computer vision.' At best, detection/tracking is one component in an autonomy stack, and the paper needs either to reframe its claim as 'object detection and tracking are useful components' or to provide an explicit integration argument showing how these components alone achieve autonomous navigation.
  3. [Section 2] The motivation statement that using computer vision 'reduces dependency on external modules like sensors and boards' is unsupported. No system can navigate with a camera alone without state estimation (e.g., IMU, visual odometry, or GPS) and a control/planning layer, and none of the reviewed papers demonstrates such a sensor-free autonomous drone. This claim is load-bearing because it is the stated justification for the entire survey.
  4. [Section 3 and Table 1] Reported accuracies (50%, 86%, 90%, 90.4%, 92%, 97.3%) are copied without error bars, dataset sizes, evaluation protocols, or hardware conditions, and the table's 'Architecture' and 'Dataset' columns are inconsistent (e.g., 'LSTM (Shallow Network)' for [11] and 'Sourced manually' for many entries). Without a defined search strategy, inclusion criteria, and critical comparison, the survey cannot support any general conclusion about accuracy, efficiency, or robustness.
minor comments (5)
  1. [Abstract and Section 1] There are numerous typographical errors, including '21 st century', 'UA V', 'autonomous UA V is called a drone', and 'by through various algorithms' in the Abstract.
  2. [Section 3] 'There proposed system' should be 'Their proposed system', and 'Kernalized' should be 'Kernelized' in the description of [26].
  3. [Table 1] The row for Kwak et al. [23] states that the method uses A* 'instead of computer vision', which contradicts the paper's framing of the survey as being about computer-vision-based navigation; this discrepancy needs clarification.
  4. [Section 3] Some summaries are not tied to precise experimental settings; for example, the '50% when detecting an object and 86% when classifying' result for [9] lacks the definition of detection vs. classification and the test set, so the number as reported is not interpretable.
  5. [References] Many entries are incomplete or inconsistently formatted (e.g., missing page ranges for [16], partial volume/issue fields for [5], [7], and [8]); please align all references with the journal's style.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular reasoning: the paper is a literature survey with no derivation chain, fitted parameters, or self-cited load-bearing claims.

full rationale

The manuscript is explicitly a compendium and literature survey: its stated purpose is to 'review the various approaches several authors have proposed for the purpose of autonomous navigation of UAVs by through various algorithms of object detection and tracking in real time.' It constructs no mathematical derivation, makes no quantitative prediction, fits no parameters, and introduces no novel formalism whose output is equivalent to its input by construction. The central assertion that UAV challenges 'can be solved by programming UAVs to be autonomous, using object detection and tracking, through Computer Vision algorithms' is a general motivating claim, not a derived result; it is unsupported by the survey and potentially inconsistent with cited approaches, but unsupported claims and internal inconsistency are correctness risks, not circularity. The paper's tables and text summarize external prior work (e.g., Valenti et al., Maciel-Pearson et al., Kwak et al.), and none of these summaries is used as both premise and conclusion. There are no self-citations that carry the argument, no imported uniqueness theorems from the authors' own prior work, and no renamed known result presented as a derivation. Therefore the appropriate circularity finding is none, with score 0.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

The paper contributes no free parameters or invented entities. It relies on domain assumptions: that vision algorithms can run in real time on UAV hardware, that vision alone can replace external sensors and GPS, and that the accuracy values quoted from the cited papers are trustworthy. None of these is demonstrated in this text.

assumptions (3)
  • domain assumption Object detection and tracking algorithms can run in real time on UAV onboard hardware.
    Section 1 asserts algorithms can be developed suitable to hardware so as to allow real time processing, but no onboard benchmark is provided.
  • domain assumption Computer vision alone can address the listed UAV challenges, including signal quality, range, human expertise, hardware robustness, and data security, and can reduce dependency on external sensors.
    Section 2 Motivation claims this reduces dependency on external modules; the statement is asserted, not demonstrated.
  • domain assumption The accuracy figures reported in the cited works are accurate and comparable across different datasets and hardware.
    Table 1 lists accuracies such as 97.3% without citing dataset splits, error bars, or measurement procedures.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Compendium of Autonomous Navigation using Object Detection and Tracking in Unmanned Aerial Vehicles." pith.science (2026). https://pith.science/paper/LBWHP5EB

@misc{pith2026250605378,
  author       = {Pith},
  title        = {Pith review of: A Compendium of Autonomous Navigation using Object Detection and Tracking in Unmanned Aerial Vehicles},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LBWHP5EB}},
  note         = {Machine review of arXiv:2506.05378}
}
read the original abstract

Unmanned Aerial Vehicles (UAVs) are one of the most revolutionary inventions of 21st century. At the core of a UAV lies the central processing system that uses wireless signals to control their movement. The most popular UAVs are quadcopters that use a set of four motors, arranged as two on either side with opposite spin. An autonomous UAV is called a drone. Drones have been in service in the US army since the 90's for covert missions critical to national security. It would not be wrong to claim that drones make up an integral part of the national security and provide the most valuable service during surveillance operations. While UAVs are controlled using wireless signals, there reside some challenges that disrupt the operation of such vehicles such as signal quality and range, real time processing, human expertise, robust hardware and data security. These challenges can be solved by programming UAVs to be autonomous, using object detection and tracking, through Computer Vision algorithms. Computer Vision is an interdisciplinary field that seeks the use of deep learning to gain a high-level understanding of digital images and videos for the purpose of automating the task of human visual system. Using computer vision, algorithms for detecting and tracking various objects can be developed suitable to the hardware so as to allow real time processing for immediate judgement. This paper attempts to review the various approaches several authors have proposed for the purpose of autonomous navigation of UAVs by through various algorithms of object detection and tracking in real time, for the purpose of applications in various fields such as disaster management, dense area exploration, traffic vehicle surveillance etc.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

27 extracted references · 27 canonical work pages

  1. [20]

    Enabling Computer Vision-Based Autonomous Navigation for Unmanned Aerial Vehicles in Cluttered GPS-Denied Environments,

    F. Valenti, D. Giaquinto, L. Musto, A. Zinelli, M. Bertozzi, and A. Broggi, “Enabling Computer Vision-Based Autonomous Navigation for Unmanned Aerial Vehicles in Cluttered GPS-Denied Environments,” IEEE Conf. Intell. Transp. Syst. Proceedings, ITSC, vol. 2018-Novem, pp. 3886–3891, 2018

    work page 2018

  2. [11]

    Multi-Task Regression-based Learning for Autonomous Unmanned Aerial Vehicle Flight Control within Unstructured Outdoor Environments,

    B. G. Maciel Pearson, S. Akcay, A. Atapour-Abarghouei, C. Holder, and T. Breckon, “Multi-Task Regression-based Learning for Autonomous Unmanned Aerial Vehicle Flight Control within Unstructured Outdoor Environments,” IEEE Robot. Autom. Lett., vol. 4, no. 4, p. 8, 2019

    work page 2019

  3. [23]

    Autonomous UAV Flight Control for GPS-Based Navigation,

    J. Kwak and Y. Sung, “Autonomous UAV Flight Control for GPS-Based Navigation,” IEEE Access, vol. 6, no. c, pp. 37947–37955, 2018

    work page 2018

  4. [1]

    Detection, Tracking, and Interdiction for Amateur Drones,

    I. Guvenc, F. Koohifar, S. Singh, M. L. Sichitiu, and D. Matolak, “Detection, Tracking, and Interdiction for Amateur Drones,” IEEE Commun. Mag., vol. 56, no. 4, pp. 75–81, 2018

    work page 2018

  5. [2]

    VisDrone-VDT2018: The Vision Meets Drone Video Detection and Tracking Challenge Results: Munich, Germany, September 8-14, 2018, Proceedings, Part V,

    C. R. Munich et al., “VisDrone-VDT2018: The Vision Meets Drone Video Detection and Tracking Challenge Results: Munich, Germany, September 8-14, 2018, Proceedings, Part V,” vol. 11206, no. January, pp. 1–23, 2018

    work page 2018

  6. [3]

    Vision Meets Drones: A Challenge,

    P. Zhu, L. Wen, X. Bian, H. Ling, and Q. Hu, “Vision Meets Drones: A Challenge,” pp. 1–11, 2018

    work page 2018

  7. [4]

    Computer Vision – ECCV 2018 Workshops,

    S. C. B, N. Koznek, A. Ismail, G. Adam, V. Narayan, and M. Schulze, “Computer Vision – ECCV 2018 Workshops,” in European Conference on Computer Vision 2018, 2019, vol. 11133, p. 23

    work page 2018

  8. [5]

    Real-Time Embedded Computer Vision on UAVs,

    K. Van Beeck, T. Tuytelaars, D. Scarramuza, and T. Goedemé, “Real-Time Embedded Computer Vision on UAVs,” vol. 1, p. 8, 2019

    work page 2019

Show all 27 references
  1. [6]

    Vision-Based Real-Time Aerial Object Localization and Tracking for UAV Sensing System,

    Y. Wu, Y. Sui, and G. Wang, “Vision-Based Real-Time Aerial Object Localization and Tracking for UAV Sensing System,” IEEE Access, vol. 5, pp. 23969–23978, 2017

    2017

  2. [7]

    Development of an autonomous UAV platform for advanced research applications,

    H. B. Kurt and E. Altuǧ, “Development of an autonomous UAV platform for advanced research applications,” ACM Int. Conf. Proceeding Ser., pp. 29–32, 2017

    2017

  3. [8]

    The object detection based on deep learning,

    C. Tang, Y. Feng, X. Yang, C. Zheng, and Y. Zhou, “The object detection based on deep learning,” in Proceedings - 2017 4th International Conference on Information Science and Control Engineering, ICISCE 2017, 2017, pp. 723–728

    2017

  4. [9]

    Seeing Eye Drone: A Deep Learning, Vision-based UAV for Assisting the Visually Impaired with Mobility,

    L. Grewe and G. Stevenson, “Seeing Eye Drone: A Deep Learning, Vision-based UAV for Assisting the Visually Impaired with Mobility,” Proc. ACM Turing Celebr. Conf. - China, p. 5, 2019

    2019

  5. [10]

    Drone services: issues in drones for location-based services from human-drone interaction to information processing,

    M. Alwateer, S. W. Loke, and A. M. Zuchowicz, “Drone services: issues in drones for location-based services from human-drone interaction to information processing,” J. Locat. Based Serv., vol. 13, no. 2, p. 35, 2019

    2019

  6. [12]

    Intelligent vision system for multi-rotor UAVs in SAR operation,

    T. T. Ngo and D. S. Kim, “Intelligent vision system for multi-rotor UAVs in SAR operation,” ACM Int. Conf. Proceeding Ser., vol. Part F1479, p. 4, 2019

    2019

  7. [13]

    Unmanned Aerial Vehicle Video-based Target Tracking Algorithm Using Sparse Representation,

    M. Wan, G. Gu, W. Qian, K. Ren, X. Maldague, and Q. Chen, “Unmanned Aerial Vehicle Video-based Target Tracking Algorithm Using Sparse Representation,” IEEE Internet Things J., vol. PP, no. c, p. 17, 2019

    2019

  8. [14]

    Effective Contexts for UAV Vehicle Detection,

    J. Yang, X. Xie, and W. Yang, “Effective Contexts for UAV Vehicle Detection,” IEEE Access, vol. 7, p. 13, 2019

    2019

  9. [15]

    Correction of Barrel Distortion in Fisheye Lens Images Using Image-Based Estimation of Distortion Parameters,

    M. Lee, H. Kim, and J. Paik, “Correction of Barrel Distortion in Fisheye Lens Images Using Image-Based Estimation of Distortion Parameters,” IEEE Access, vol. 7, pp. 45723–45733, 2019

    2019

  10. [16]

    AR-drone as a platform for robotic research and education,

    T. Krajník, V. Vonásek, D. Fišer, and J. Faigl, “AR-drone as a platform for robotic research and education,” Commun. Comput. Inf. Sci., vol. 161 CCIS, pp. 172–186, 2011

    2011

  11. [17]

    Human Object Detection in Forest with Deep Learning based on Drone’s Vision,

    S. P. Yong and Y. C. Yeong, “Human Object Detection in Forest with Deep Learning based on Drone’s Vision,” in 2018 4th International Conference on Computer and Information Sciences: Revolutionising Digital Landscape for Sustainable Smart Society, ICCOINS 2018 - Proceedings, 20...

    2018

  12. [18]

    Drone-Augmented human vision: Exocentric control for drones exploring hidden areas,

    O. Erat, W. A. Isop, D. Kalkofen, and D. Schmalstieg, “Drone-Augmented human vision: Exocentric control for drones exploring hidden areas,” IEEE Trans. Vis. Comput. Graph., vol. 24, no. 4, p. 10, 2018

    2018

  13. [19]

    Hand gesture controlled drones: An open source library,

    K. Natarajan, T. H. D. Nguyen, and M. Mete, “Hand gesture controlled drones: An open source library,” Proc. - 2018 1st Int. Conf. Data Intell. Secur. ICDIS 2018, pp. 168–175, 2018

    2018

  14. [21]

    Learning to Detect Aircraft for Long-Range Vision-Based Sense-and-Avoid Systems,

    J. James, J. J. Ford, and T. L. Molloy, “Learning to Detect Aircraft for Long-Range Vision-Based Sense-and-Avoid Systems,” IEEE Robot. Autom. Lett., vol. 3, no. 4, pp. 4383–4390, 2018

    2018

  15. [22]

    Coarse-to-Fine UAV Target Tracking With Deep Reinforcement Learning,

    W. Zhang, K. Song, X. Rong, and Y. Li, “Coarse-to-Fine UAV Target Tracking With Deep Reinforcement Learning,” IEEE Trans. Autom. Sci. Eng., vol. PP, pp. 1–9, 2018

    2018

  16. [24]

    Object tracking with multi-view support vector machines,

    S. Zhang, X. Yu, Y. Sui, S. Zhao, and L. Zhang, “Object tracking with multi-view support vector machines,” IEEE Trans. Multimed., vol. 17, no. 3, pp. 265–278, 2015

    2015

  17. [25]

    Graph-Regularized Structured Support Vector Machine for Object Tracking,

    S. Zhang, Y. Sui, S. Zhao, and L. Zhang, “Graph-Regularized Structured Support Vector Machine for Object Tracking,” IEEE Trans. Circuits Syst. Video Technol., vol. 27, no. 6, pp. 1249–1262, 2017

    2017

  18. [26]

    Real-Time Obstacle Detection and Tracking for Sense-and-Avoid Mechanism in UAVs,

    S. P. Bharati, Y. Wu, Y. Sui, C. Padgett, and G. Wang, “Real-Time Obstacle Detection and Tracking for Sense-and-Avoid Mechanism in UAVs,” IEEE Trans. Intell. Veh., vol. 3, no. 2, pp. 185–197, 2018

    2018

  19. [27]

    Urban Traffic Density Estimation Based on Ultrahigh-Resolution UAV Video and Deep Neural Network,

    J. Zhu et al., “Urban Traffic Density Estimation Based on Ultrahigh-Resolution UAV Video and Deep Neural Network,” IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens., vol. 11, no. 12, pp. 4968–4981, 2018

    2018

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.