Table 5 Frequency distribution of video frames in the UNBC-McMaster and DISFA databases by PSPI scores and PSPI groups.

From: Automated detection of pain levels using deep feature extraction from shutter blinds-based dynamic-sized horizontal patches with facial images

PSPI score

UNBC-McMaster database

DISFA database

Frequency by PSPI score

Frequency by PSPI group

Frequency by PSPI score

Frequency by PSPI group

0

40,029

2483*

90,309

9025*

1

2909

2909

9973

9973

2

2351

3763

12,194

10,309*

3

1412

8447

4

802

1697

4127

9875

5

242

1661

6

270

1526

7

53

996

8

79

435

9

32

503

10

67

317

11

76

169

12

48

108

13

22

33

14

1

0

15

5

0

Total

48,398

10,852

130,798

39,182

  1. *Random under-sampling of over-represented PSPI groups was performed to create more balanced study datasets with smaller total numbers of video frames for training and testing the pain intensity classification model.
Back to article page