CoChat

A Systematic Review of Drone Based Road Traffic Monitoring System

Open access

The DOI matches its Crossref record; 153 of the 153 references it cites were found in a registry or an index; 37 of the 184 sources its claims cite were checked for support.

Abstract

Drone deployment has become crucial in a variety of applications, including solutions to traffic issues in metropolitan areas and highways. On the other hand, data collected via drones suffers from several problems, including a wide range of object scales, angle variations, truncation, and occlusion. To process and manipulate visual data from the drones, a variety of image processing algorithms have been employed, each with a distinct aim. Additionally, recent breakthroughs in the field of Artificial Intelligence, particularly deep learning, have attracted broad interest and are being applied to many domains in the framework of smart cities, including road traffic monitoring. The purpose of this study is to conduct a systematic review of drone-based traffic monitoring systems from a deep learning perspective. This work focuses on vehicle detection, tracking, and counting, since they are fundamental building blocks towards founding solutions for traffic congestion, flow rate and vehicle speed estimation. Additionally, drone-based datasets are examined, which face issues and problems caused by the diversity of features inherent of drone devices. The review analysis presented in this work summarizes the literature solutions provided and deployed so far and discusses future research trends in establishing a comprehensive traffic monitoring system in support of the development of smart cities.

Overview

CoChat Summary

TL;DR

This systematic review surveys deep-learning approaches for drone-based road traffic monitoring, focusing on vehicle detection, tracking, and counting, plus aerial datasets and evaluation metrics. It finds that custom datasets often yield >80% performance while public datasets such as VisDrone and DOTA show lower results, and identifies the lack of a large-scale diverse drone traffic dataset as a key open issue.

Key findings

  • The review screened 37 journals, 52 conference papers, and 9 other items, including 19 journals and 13 conference papers, for the publication period 2015–2021; the shortlisted articles total 32 across detection, counting, and tracking (Tables 1–2).
  • Most systems use one- or two-stage detectors, with Faster R-CNN and YOLOv3 most widely employed; custom datasets often achieve >80% performance, while public datasets such as VisDrone and DOTA yield lower performance.
  • Counting accuracy exceeds 90% with custom datasets (e.g., 93.7% counting-by-tracking [3], 97.58% daytime lane-based counting [76]), and tracking performance is >80% in the majority of studies (e.g., MOSSE tracker MOTA 90.91%, MOTP 92.11% [5]).
  • Nighttime detection/counting is lower than daytime (e.g., 84.6% night vs 87.9% day detection [76]; 85.26% night vs 97.58% day counting [76]), and a drone-based large-scale diverse aerial dataset comparable to MS-COCO is still missing.

Methods

  • Systematic review with research questions on datasets/metrics (RQ1), DL frameworks and pre-processing for detection/tracking/counting (RQ2), and effectiveness/challenges/future trends (RQ3); two-step screening by title/abstract then full-text reading, plus reference scanning.
  • Inclusion: English-language publications from 2015–2021; journals, conferences, and technical reports were screened, with final inclusion of 19 journal and 13 conference articles (Table 1).
  • Analysis organizes literature into image-based detection, video-based detection, on-board monitoring, pre-processing, counting, and tracking, with comparison tables of frameworks, datasets, metrics, and outcomes.

Limitations

  • Paper-stated: exact UAV altitude is not always reported; public datasets yield lower performance; custom datasets often use limited object classes (up to four target classes in many systems); nighttime performance is lower; tracking outcomes are not directly comparable because systems use different datasets and class counts.
  • Paper-stated: a drone-based large-scale diverse aerial dataset comparable to MS-COCO is still lacking; on-board processing is computationally costly in power consumption, flight duration, and battery life; privacy concerns can force higher altitudes or side views, reducing performance.
  • Evident from text: the review is restricted to 2015–2021 English-language literature and focuses solely on detection, tracking, and counting, excluding congestion analysis, flow rate, and speed estimation despite noting them as traffic monitoring tasks.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1109/access.2022.3207282

Cites

153 of 153 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

15 Valid · 21 Not valid · 122 Not checked · 69 Own results and interpretations

93 more claims · 9 with their cited sources checked

Create a free account to see all 227 claims and to verify the 48 references still open.

See all claims — free account
  • It may state two separate claims as one. 59
  • Probably not a substantial claim. 95
  • The quote only partly states this claim. 1

Create a free account to see these claims.

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

4 of 4

Citations

Cites

Research works printed but not linked: 7

  1. Masood, 2017, Trajectory tracking and hurdle avoidance for an autonomous quadrotor UAV, Eur. J. Adv. Eng. Technol., 4, 135
  2. Kaiming, 2017, Mask R-CNN, IEEE Trans. Pattern Anal. Mach. Intell., 42, 386
  3. Park, Large Scale Visual Recognition Challenge 2017
  4. 10.1109/TPAMI.2016.2577031 · DOI · Paper page
  5. Collins, An open source tracking testbed and evaluation web site, 2, 35
  6. Glenn, 2021, YOLOv5
  7. 2022, YOLOv6

Cited by

See all 139

Metadata

Authors
Bisio, I., Garibotto, C., Haleem, H., Lavagetto, F., & Sciarrone, A.
Affiliations
University of Genoa (IT)
Venue
IEEE Access
Published
2022
Volume
10
Pages
101537-101555
ISSN
2169-3536 (ISSN-L)
OpenAlex
W4296211548
Cited by
139
References
148
Citations by year
2026: 41 · 2025: 49 · 2024: 27 · 2023: 21 · 2022: 1
FWCI
1.76
Type
journal article
Language
en
Open access
Locations
  • IEEE Access (publishedVersion, cc-by)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • CINECA IRIS Institutial Research Information System (University of Genoa) (submittedVersion)
Topics
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Video Surveillance and Tracking Methods
  • Physical Sciences › Engineering › Automotive Engineering › Autonomous Vehicle Technology and Safety
Keywords
unmanned aerial vehicles, vehicle detection, vehicle tracking, vehicle counting, traffic congestion, traffic flow estimation, vehicle speed estimation, smart cities, deep learning

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol and qwen3.8-omni-flash).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.