Skip to main content

Table 3 Average classification errors (≈ EERs) resulting from our leave-one-subject-out cross validation

From: Estimation of gait normality index based on point clouds through deep auto-encoder

Model

Training data

Data type

Classification error (leave-one-out)

   

Per-frame

Segment

Entire seq.

HMM [21]

Normal only

Skeleton

–

0.396

0.198

One-class SVM [5]

Normal only

Silhouette

0.418

0.274

0.136

Binary SVM [5]

Normal + abnormal

Silhouette

0.110

0.152

0.111

HMM [23]

Normal only

Depth map

–

0.473

0.431

Cross-correlation [23]

Normal only

Silhouette

–

0.321

0.097

HMM + cross-correlation [23]

Normal only

Silhouette + depth map

–

0.319

0.083

(Our) Sigmoid

Normal only

Point cloud

0.362

0.240

0.160

(Our) Sigmoid + dropout

Normal only

Point cloud

0.363

0.241

0.148

(Our) Tanh

Normal only

Point cloud

0.298

0.144

0.049

(Our) Tanh + dropout

Normal only

Point cloud

0.301

0.168

0.074

(Our) Leaky ReLU

Normal only

Point cloud

0.297

0.173

0.099

(Our) Leaky ReLU + dropout

Normal only

Point cloud

0.311

0.185

0.123

(Our) Multi-network

Normal only

Point cloud

0.303

0.178

0.086

  1. The italic values indicate the best results in different evaluations