Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Video

Updated: Jan 9, 2026

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
06:37

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention

Published on: December 15, 2023

5.2K

Multi-stage classification of abnormal traffic events using a multi-head + LSTM.

Pratik Jadhav1,2, Abderrahim Benslimane3, Deepali R Vora4

  • 1Artificial Intelligence and Machine Learning Department, Symbiosis Institute of Technology, Symbiosis International (Deemed) University, Pune, 412115, India.

Scientific Reports
|December 8, 2025
PubMed
Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Design and simulation of a hybrid PLC-AI automotive control module for adaptive lighting and honking.

Scientific reports·2026
Same author

An ensemble of deep learning models with falcon optimization assisted diabetic retinopathy diagnosis on retinal fundus images.

Scientific reports·2026
Same author

Kinematic joint angles do not predict energy expenditure index in sub-acute and chronic stroke survivors: An exploratory study.

Journal of bodywork and movement therapies·2026
Same author

Sleep Disturbance and its Relationship with Behavioral Problems in Children with Down's Syndrome - A Cross-Sectional Study.

Indian journal of pediatrics·2026
Same author

Rose leaf disease classification and severity estimation using an interpretable vision transformer-based multi-task framework.

BMC plant biology·2026
Same author

Ensuring the integrity of AI models: a blockchain-based approach for protecting medical imaging training data.

Scientific reports·2026

This study introduces a novel multi-head+LSTM model for advanced traffic analysis. The model accurately detects anomalies, classifies congestion, and identifies incidents, improving urban transportation management.

Area of Science:

  • Intelligent Transportation Systems (ITS)
  • Deep Learning Applications
  • Urban Mobility Analysis

Background:

  • Traffic congestion, anomalies, and incidents critically affect urban transportation efficiency and road safety.
  • Existing statistical and machine learning models struggle with the dynamic nature of traffic patterns.
  • Accurate event detection and classification are vital for traffic management, emergency response, and infrastructure planning.

Purpose of the Study:

  • To develop a robust framework for detecting anomalies, classifying congestion levels, and identifying traffic incidents.
  • To enhance the accuracy of traffic event analysis by integrating temporal dependencies and contextual weather information.
  • To overcome the generalization limitations of traditional models in complex, evolving traffic environments.
Keywords:
Anomaly detectionCongestion classificationIncident identificationIntelligent transportation systems (ITS)LSTMMulti-head attentionTraffic flow prediction

Related Experiment Videos

Last Updated: Jan 9, 2026

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention
06:37

Author Spotlight: Addressing Technical and Subjective Challenges in Measuring Classroom Attention

Published on: December 15, 2023

5.2K

Main Methods:

  • Proposed a multi-head+LSTM model within a multistage classification framework.
  • Employed Isolation Forest for anomaly detection, K-means clustering for congestion classification (low, medium, high), and a spatial threshold (1.5 km) for incident identification.
  • Trained the model on 15 days of PeMS traffic data, incorporating weather information for improved predictive accuracy.

Main Results:

  • The proposed multi-head+LSTM model significantly outperformed existing methods across all classification stages.
  • Achieved higher precision, recall, F1-score, and ROC-AUC compared to traditional approaches.
  • Demonstrated robust performance in anomaly detection, congestion classification, and incident identification.

Conclusions:

  • The deep learning-based approach offers a significant advancement for intelligent transportation systems.
  • The model's ability to capture temporal dependencies and integrate weather data enhances traffic analysis robustness.
  • This framework enables data-driven decision-making for more effective urban traffic management and improved road safety.