Multimodal Transformer-Based Driver Emotion and Behavior Analysis: Supporting Code and Dataset
Year: 2026
Type: software
Abstract: Multimodal Transformer-Based Driver Emotion and Behavior Analysis: Research Code and Dataset This repository provides research code, a multimodal driver emotion and behavior dataset, and supporting documentation associated with the study entitled “A Multimodal Transformer-Based Framework for Driver Emotion and Behavior Analysis in Complex Traffic Scenarios.” The dataset contains 1,000 driver-contextual samples covering five emotional states: Calm, Stress, Anger, Fear, and Fatigue. It includes numerical features representing facial stress indicators, gaze orientation, physiological measurements (heart rate, heart rate variability, electrodermal activity, and EEG beta power), traffic-context variables, and driver behavioral performance measures. The accompanying Python implementation uses PyTorch to construct a transformer-based emotion classifier incorporating feature projection, transformer encoder layers, and a classification head. The evaluation script implements five-fold stratified cross-validation and calculates classification accuracy, precision, recall, and weighted F1-score. Repository Contents Multimodal driver emotion and behavior dataset in CSV format. PyTorch transformer training and evaluation code. Python dependency specifications. README documentation and supporting research materials. Research Purpose The deposited resources support research transparency, methodological assessment, and further development of multimodal driver emotion recognition and intelligent transportation systems. Data Provenance and Implementation Scope The provided implementation covers the framework described in the manuscript. Keywords: Driver Emotion Recognition; Multimodal Learning; Transformer; Driver Behavior Analysis; Deep Learning; PyTorch; Intelligent Transportation Systems.
Author: Khan, Dr. Sajid Ullah
Institution: Lahore Garrison University
Language: English
Topic: Emotion and Mood Recognition
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7225083138