Table 2.

Evaluations of all our methods.

Model	Feature	ICC (95% CI)	τ b (95% CI)	P K (95% CI)
GBDT with Huber loss	VGG	0.91 [0.87, 0.93]	0.78 [0.69, 0.79]	0.90 [0.86, 0.91]
	ResNet	0.91 [0.87, 0.93]	0.77 [0.73, 0.81]	0.90 [0.87, 0.91]
	Inception	0.87 [0.84, 0.90]	0.71 [0.65, 0.76]	0.86 [0.84, 0.89]
GBDT with RankNet	VGG	0.91 [0.88, 0.93]	0.77 [0.72, 0.81]	0.89 [0.87, 0.91]
	ResNet	0.91 [0.88, 0.94]	0.79 [0.75, 0.82]	0.90 [0.88, 0.92]
	Inception	0.88 [0.85, 0.91]	0.74 [0.70, 0.79]	0.88 [0.86, 0.90]
SVR	VGG	0.94 [0.92, 0.95]	0.80 [0.76, 0.84]	0.91 [0.89, 0.93]
	ResNet	0.95 [0.93, 0.96]	0.83 [0.79, 0.86]	0.93 [0.91, 0.94]
	Inception	0.89 [0.85, 0.92]	0.74, [0.69, 0.79]	0.88 [0.85, 0.90]
Ranking SVM	VGG	0.91 [0.88, 0.93]	0.76 [0.71, 0.80]	0.89 [0.87, 0.91]
	ResNet	0.92 [0.89, 0.94]	0.78 [0.73, 0.82]	0.90 [0.88, 0.92]
	Inception	0.89 [0.86, 0.92]	0.76 [0.71, 0.80]	0.89 [0.86, 0.91]

Bold indicates the best performance in terms of the corresponding metric.