CoChat

A Real-Time Computer Vision Based Approach to Detection and Classification of Traffic Incidents

Open access
Basheer Ahmed, Mohammed Imran, Zaghdoud, Rim, Ahmed, Mohammed Salih, Sendi, Razan, Alsharif, Sarah, Alabdulkarim, Jomana, Albin Saad, Bashayr Adnan, Alsabt, Reema, Rahman, Atta, Krishnasamy, Gomathi · 2023 · Big Data and Cognitive Computing

The DOI matches its Crossref record; 29 of the 29 references it cites were found in a registry or an index; 6 of the 19 sources its claims cite were checked for support.

Abstract

To constructively ameliorate and enhance traffic safety measures in Saudi Arabia, a prolific number of AI (Artificial Intelligence) traffic surveillance technologies have emerged, including Saher, throughout the past years. However, rapidly detecting a vehicle incident can play a cardinal role in ameliorating the response speed of incident management, which in turn minimizes road injuries that have been induced by the accident’s occurrence. To attain a permeating effect in increasing the entailed demand for road traffic security and safety, this paper presents a real-time traffic incident detection and alert system that is based on a computer vision approach. The proposed framework consists of three models, each of which is integrated within a prototype interface to fully visualize the system’s overall architecture. To begin, the vehicle detection and tracking model utilized the YOLOv5 object detector with the DeepSORT tracker to detect and track the vehicles’ movements by allocating a unique identification number (ID) to each vehicle. This model attained a mean average precision (mAP) of 99.2%. Second, a traffic accident and severity classification model attained a mAP of 83.3% while utilizing the YOLOv5 algorithm to accurately detect and classify an accident’s severity level, sending an immediate alert message to the nearest hospital if a severe accident has taken place. Finally, the ResNet152 algorithm was utilized to detect the ignition of a fire following the accident’s occurrence; this model achieved an accuracy rate of 98.9%, with an automated alert being sent to the fire station if this perilous event occurred. This study employed an innovative parallel computing technique for reducing the overall complexity and inference time of the AI-based system to run the proposed system in a concurrent and parallel manner.

Overview

CoChat Summary

TL;DR

The paper presents a real-time computer vision system for detecting traffic incidents, classifying accident severity, and detecting postcollision vehicle fires, with automated email alerts. It integrates YOLOv5+DeepSORT, YOLOv5 severity classification, and ResNet152 fire detection into a GUI using parallel threads.

Key findings

  • Vehicle detection and tracking: YOLOv5+DeepSORT achieved mAP 99.2%, precision 98.4%, and recall 97.5%; it was selected over YOLOR (mAP 99.4%, precision 90.3%, recall 98.6%) and YOLOv4-Tiny (mAP 91.71%, precision 93.1%, recall 91%, F1-score 90.5%).
  • Accident severity classification: YOLOv5 achieved mAP 83.3%, precision 84.01%, and recall 76.7%; comparison algorithms were YOLOR (mAP 87.2%, precision 73.9%, recall 86.2%) and Faster R-CNN (mAP 79.8%, recall 68.6%).
  • Postcollision fire detection: ResNet152 achieved accuracy 98.955%, precision 98.72%, and recall 98.905%, outperforming InceptionV3 (accuracy 97.563%, precision 98.178%, recall 96.768%).
  • The system sends email alerts to the nearest medical center for severe accidents and to the fire station for fires, with parallel QThread integration intended to reduce inference time.

Methods

  • Three-model pipeline: vehicle detection/tracking (YOLOv5 + DeepSORT), accident severity classification (YOLOv5, YOLOR, Faster R-CNN compared), and postcollision fire detection (ResNet152, InceptionV3 compared), plus an automated email alert system and QT-based GUI integration.
  • Data: a custom vehicle dataset of 8000 annotated images containing approximately 36,400 vehicle instances from CCTV; an accident dataset of 5185 manually annotated frames labeled moderate or severe, combining online/CCTV/YouTube sources and semirealistic video game videos; and a fire dataset of 2750 fire images and 3200 no-fire images from YouTube. Splits were 80/10/10 for training, validation, and testing.
  • Preprocessing and training: YOLO models resized to 416×416; ResNet152 to 256×256 and InceptionV3 to 299×299; augmentation included brightness changes of ±25%, Gaussian blur with kernel size 1.75 pixels for vehicle data, and random resized crop, center crop, random horizontal flip, and normalization for fire data. Training used Google Colab’s virtual machine and an NVIDIA DGX Personal AI Supercomputer, with cross-entropy loss, Adam optimizer, and early stopping.

Limitations

  • Stated: computation time for running three computer vision models on live or captured video is a bottleneck, addressed with concurrent thread-based parallelization; future work includes license plate localization, expanding the dataset, increasing epochs, and deploying on real CCTV in high-incident Saudi regions.
  • Evident: the accident severity model has lower mAP (83.3%) and recall (76.7%) than the vehicle and fire models; the accident dataset includes semirealistic video game videos, which may not fully represent real incidents; and evaluation is on custom datasets and a prototype GUI rather than real-world CCTV deployment, which the paper lists as future work.

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

Integrity signals

Registry

Crossref · 10.3390/bdcc7010022

Cites

29 of 29 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Figure 1 Methodological workflow of the proposed system. Figure 1. Methodological workflow of the proposed system.
Figure 1 Methodological workflow of the proposed system. Figure 1. Methodological workflow of the proposed system.From the PDF
Figure 2 Phase 1 Flowchart for the Detection and Tracking of Vehicles. Figure 2. Phase 1 Flowchart for the Detection and Tracking of Vehicles.
Figure 2 Phase 1 Flowchart for the Detection and Tracking of Vehicles. Figure 2. Phase 1 Flowchart for the Detection and Tracking of Vehicles.From the PDF
Figure 3 Image Sampled from the Vehicle Dataset: ( a ) Nighttime Images, ( b ) Daylight Images, ( c ) Sandstorm Images. Figure 3. Image Sampled from the Vehicle Dataset: ( a ) Nighttime Images, ( b ) Daylight Images, ( c ) Sandstorm Images.
Figure 3 Image Sampled from the Vehicle Dataset: ( a ) Nighttime Images, ( b ) Daylight Images, ( c ) Sandstorm Images. Figure 3. Image Sampled from the Vehicle Dataset: ( a ) Nighttime Images, ( b ) Daylight Images, ( c ) Sandstorm Images.From the PDF
Figure 4 Phase 2 Flowchart for the Classification of Accident Severity Level. Figure 4. Phase 2 Flowchart for the Classification of Accident Severity Level.
Figure 4 Phase 2 Flowchart for the Classification of Accident Severity Level. Figure 4. Phase 2 Flowchart for the Classification of Accident Severity Level.From the PDF
Figure 5 Phase 3 Flowchart for Detecting Post Collision Fires. Figure 5. Phase 3 Flowchart for Detecting Post Collision Fires.
Figure 5 Phase 3 Flowchart for Detecting Post Collision Fires. Figure 5. Phase 3 Flowchart for Detecting Post Collision Fires.From the PDF
Image 6
From the PDF
Image 7
From the PDF
Figure 6 Sample of the Post Collision Dataset: ( a ) displays the fire images and ( Figure 6. Sample of the Post Collision Dataset: ( a ) displays the fire images and ( b ) no-fire.
Figure 6 Sample of the Post Collision Dataset: ( a ) displays the fire images and ( Figure 6. Sample of the Post Collision Dataset: ( a ) displays the fire images and ( b ) no-fire.From the PDF
Image 9
From the PDF
Image 10
From the PDF
Image 11
From the PDF
Figure 7 Email Alert System Workflow Flowchart. Figure 7. Email Alert System Workflow Flowchart.
Figure 7 Email Alert System Workflow Flowchart. Figure 7. Email Alert System Workflow Flowchart.From the PDF
Figure 8 Proposed Systems Model Integration Flowchart. Figure 8. Proposed Systems Model Integration Flowchart.
Figure 8 Proposed Systems Model Integration Flowchart. Figure 8. Proposed Systems Model Integration Flowchart.From the PDF
Figure 9 ResNet152 Learning Curves:( a ) Accuracy, ( b ) Loss. Figure 9. ResNet152 Learning Curves:( a ) Accuracy, ( b ) Loss.
Figure 9 ResNet152 Learning Curves:( a ) Accuracy, ( b ) Loss. Figure 9. ResNet152 Learning Curves:( a ) Accuracy, ( b ) Loss.From the PDF
Figure 10 Vehicle Detection and Tracking Comparison Analysis Bar Chart. Figure 10. Vehicle Detection and Tracking Comparison Analysis Bar Chart.
Figure 10 Vehicle Detection and Tracking Comparison Analysis Bar Chart. Figure 10. Vehicle Detection and Tracking Comparison Analysis Bar Chart.From the PDF

Verification

Claims the paper makes

4 Valid · 5 Not valid · 13 Not checked · 24 Own results and interpretations

11 more claims

Create a free account to see all 46 claims and every source verdict.

See all claims — free account
  • Probably not a substantial claim. 22
  • It may state two separate claims as one. 14
  • The quote only partly states this claim. 3

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.

3 of 3

Citations

Cites

Research works printed but not linked: 6

  1. Nancy, 2020, Highway Accident Detection and Notification Usingmachine Learning, Int. J. Comput. Sci. Mob. Comput., 9, 168
  2. Rahman, A., and Basheer, M.I. (2019). Telemedicine Technologies, Academic Press.
  3. Alasadi, 2017, Review of data preprocessing techniques in data mining, J. Eng. Appl. Sci., 12, 4102
  4. Imran, M., and Alsuhaibani, S.A. (2019). Intelligent Data Analysis for Biomedical Applications, Academic Press.
  5. Prechelt, L. (1998). Neural Networks: Tricks of the Trade, Springer.
  6. Jose, 2021, Accident Detection and Warning System, Int. J. Innov. Sci. Res. Technol., 6, 1532

Cited by

See all 122

Metadata

Authors
Basheer Ahmed, M. I., Zaghdoud, R., Ahmed, M. S., Sendi, R., Alsharif, S., Alabdulkarim, J., Albin Saad, B. A., Alsabt, R., Rahman, A., & Krishnasamy, G.
Affiliations
  • Department of Computer Engineering, College of Computer Science and Information Technology, Imam Abdulrahman Bin Faisal University, P.O. Box 1982, Dammam 31441, Saudi Arabia
  • Imam Abdulrahman Bin Faisal University (SA)
Venue
Big Data and Cognitive Computing
Publisher
MDPI AG
Published
2023-01-28
Volume
7
Issue
1
Pages
22
ISSN
2504-2289 (ISSN-L)
OpenAlex
W4318478177
Cited by
122
References
28
Citations by year
2026: 22 · 2025: 31 · 2024: 45 · 2023: 24
FWCI
9.09
Type
journal article
Language
en
Open access
Locations
  • Big Data and Cognitive Computing (publishedVersion, cc-by)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Topics
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Video Surveillance and Tracking Methods
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
  • Physical Sciences › Engineering › Automotive Engineering › IoT and GPS-based Vehicle Safety Systems
Keywords
traffic incident detection, vehicle detection and tracking, YOLOv5, DeepSORT, fire detection, real-time traffic monitoring, incident management, parallel computing, Saudi Arabia

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 2026-09-28, under version 3 of the claim definition (prompt extract-v3).

A second model labelled each claim on 2026-09-28 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

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.