Driver Emotion Recognition Using Temporal Facial Image Sequences

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing emotion recognition methods in driving scenarios suffer from low accuracy due to limited information usage, particularly when the driver is alone or without passenger interaction, leading to ineffective audio-based solutions.

Innovation Solution

A method involving a sequence of temporally adjacent images of the driver's face is input into a pre-trained emotion recognition model, utilizing Hidden Markov Model, Conditional Random Field, or Recurrent Neural Network to predict subsequent emotional statuses based on prior and subsequent emotions, enhancing accuracy and efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If single images are used for emotion recognition, then the system complexity is low, but the recognition accuracy is insufficient

Engineering Contradiction:
Improveemotion recognition accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system performs preliminary actions by capturing a sequence of images before the actual emotion recognition occurs. These pre-captured images serve as input data that contains temporal information about emotional changes, enabling more accurate recognition without requiring complex real-time processing during the critical moment of emotion detection.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent transitions from analyzing single static images to analyzing sequences of images across the time dimension. By incorporating temporal information from multiple time points, the system adds a new dimension to the analysis, enabling detection of emotional transitions and changes that cannot be captured by single-frame analysis alone.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Measurement precision

If single images are used for emotion recognition, then the processing time is short, but the recognition accuracy is low

Engineering Contradiction:
Improveemotion recognition accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The system performs preliminary image capture and preprocessing before the actual recognition task. By preparing the sequence of images in advance and organizing them properly, the system reduces the computational burden during the critical recognition phase, maintaining efficiency while improving accuracy through temporal analysis.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system maintains continuous capture of facial images over time, creating an uninterrupted sequence that records emotional evolution. This continuous data collection ensures that no critical emotional transitions are missed, while the structured approach to processing this continuous stream maintains computational efficiency.

Inventive Principle:
Principle #20Continuity of useful action

3Measurement precision

If temporal continuity of emotional states is considered, then the accuracy improves, but the computational complexity increases

Engineering Contradiction:
Improveemotion recognition accuracyVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system segments the temporal sequence of images into discrete time points with specific emotional states. By dividing the continuous temporal data into manageable segments that can be individually analyzed and then integrated, the system reduces computational complexity while preserving the temporal continuity information needed for accurate recognition.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP4047561B1Method for recognizing an emotion of a driver, apparatus, device, medium and vehicle
Publication Date: 2026.04.22 BAYERISCHE MOTOREN WERKE AG
  • EP4047561B1 patent drawingFigure 1~2
  • EP4047561B1 patent drawingFigure 3~5
  • EP4047561B1 patent drawingFigure 6~7

AI summary

A method, an apparatus, a device, a medium, and a vehicle for recognizing an emotion of a driver. The method includes: obtaining a sequence of images to be recognized, wherein the sequence of images to be recognized includes at least two images to be recognized that are adjacent in time, and each image to be recognized includes a face of a driver; and inputting the sequence of images to be recognized into an emotion recognition model to obtain an sequence of emotion statuses, output by the emotion recognition model, that corresponds to the sequence of images to be recognized, wherein the emotion recognition model is pre-trained so that an nth emotion status in the sequence of emotion statuses is predicted based on at least one prior emotion status in the sequence of emotion statuses, wherein the at least one prior emotion status precedes the nth emotion status in time and includes an (n - 1)th emotion status, wherein n is an integer and 1 < n ≤ N, and N is a total number of the emotion statuses in the sequence of emotion statuses.