Table 1.

Category distributions of test data and drop outs of Azure, Face++, and FaceReader.

	Standardized data
	Neutral	Happy	Sad	Fear	Angry	Surprise	Disgust	Overall
Absolute frequency of images	178	178	178	178	178	178	178	1246
Relative frequency of images	14.3	14.3	14.3	14.3	14.3	14.2	14.2	100
	Drop out rates (percent per category)
Face++	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
Azure	0.0	0.0	0.0	0.0	0.0	0.0	0.0	0.0
FaceReader	0.6	0.0	0.0	1.1	1.1	2.2	1.2	0.88
	Non-standardized data
	Neutral	Happy	Sad	Fear	Angry	Surprise	Disgust	Overall
Absolute frequency of images	236	270	245	143	254	151	88	1387
Relative frequency of images	17.0	19.5	17.7	10.3	18.3	10.9	6.3	100
	Drop out rates (percent per category)
Face++	0.0	0.7	0.8	0.7	0.4	0.0	1.1	0.5
Azure	16.9	11.1	25.3	26.6	25.2	17.9	23.9	20.3
FaceReader	73.3	70.0	75.1	79.0	76.8	74.8	69.3	74.2

Percentages are rounded to the first decimal. The base of the percentage is the respective total of each category. Reading example: Azure did not find a face in 16.9% of the 236 neutral images of the non-standardized data and a total of 20.3% of the 1387 images dropped out because of no face detection.