Table 3.

Classification performance comparison to early works on the BioVid Heat Pain Database (Part A) in a LOSO cross-validation setting for the classification task $T_{0}vs.T_4$.

Approach	Description	Performance
Yang et al. [27]	BSIF	$65.17$
Kächele et al. [31,62]	Geometric Features	$65.55\pm 14.83$
Werner et al. [8]	Standardised Facial Action Descriptors	$\mathbf{72.40}$
Our Approach	Motion History Images	$65.17\pm 15.49$
Our Approach	Optical Flow Images	$69.11\pm 14.73$
Our Approach	Weighted Score Aggregation	$69.25\pm 17.31$

The performance metric consists of the average accuracy (in %) over the LOSO cross-validation evaluation. The best performing approach is depicted in bold and the second best approach is underlined.