Table 5 A review of existing DMS methods and sensor utilization.

From: Real-time driver monitoring system with facial landmark-based eye closure detection and head pose recognition

Author

Primary methodologies

DMS type

Datasets

Modalities used

Du et al.35

1D-CNN

Direct

UTA-RLDD36

Visual RGB

Huang et al.37

MTCNN, LSTM

NTHU-DDD38

Vijay et al.39

CNN, XGBoost

Ahmed et al.40

MTCNN, InceptionV3

Chen et al.41

MTCNN, ONET

Custom datasets

Frontal visual RGB, oblique visual RGB

Bakker et al.42

MTCNN, CE-CLM,

AU-CNN, DLT-LM

IN-visual RGB, out-visual RGB

Lollett et al.43

GRU-based

Visual RGB

Sharak et al.44

RFC, XGB

Visual RGB, NIR, thermal

Ansari et al.45

LSTM

Motion capture

Tran et al.46

CNN, VGG, ResNet

Two visual RGB

Awais et al.47

SVM

Four visual RGB, ECG

Lin et al.48

SVR, MLPNN, RBFNN,

Five-layer SONFIN

EOG, ECG, EEG

Massoz et al49

PDM, CLM, RLMS

ECG, EOG, EEG, EMG, Visual RGB, NIR, depth map

Kiashari et al50

ORD

NIR, thermal

Quddus et al51

LSTM

EEG, two NIR

Du et al52

MPF

Hybrid

Visual RGB, vehicle telemetry, audio, driving simulator

Kundinger et al53

SMOTE

Visual RGB, ECG, pulse, physioloigca, driving simulator

Baccour et al54

SVM

Visual RGB, driving simulator

Longxi Luo55

DSN

Visual RGB, driving simulator

Ours

YOLO v7, dlib

Direct

Custom datasets

IR