Skip to main content

Table 6 Evaluation of computational cost of each pose proposals

From: Robust head pose estimation based on key frames for human-machine interaction

Method

Time (ms per frame)

Architecture

Kazemi and Sullivan [6]*

12.0±1

CPU

Fanelli et al. [4]*

20.1±2

CPU

Li et al. [10]*

62.4±5

CPU

Li Nose*

64.2±5

CPU

KFv3*

89.9±10

CPU

Papazov et al. [12] +

122

CPU

Yang et al. [28] +

∼100

CPU

Yu et al. [9] +

∼250

CPU

Ahn et al. [24] +

0.98

GPU

Venturelli et al. [26] +

10

GPU

Saeed and Al-Hamadi [27] +

>45

GPU

  1. Results with the best performances are presented in bold
  2. *Time that we calculated
  3. +Results taken from author’s papers