Table 3.

Confusion matrix (A) and outcomes (B) of performance test for classification of three labels of segmentation-free images after training with all segmented images.

(A) Confusion Matrix		Labels
		Shallow	Proper	Deep	Sum
Prediction	Shallow	33	22	7	62
	Proper	16	127	30	173
	Deep	10	77	279	366
	Sum	59	226	316	601
(B) Outcomes		Labels
		Shallow	Proper	Deep	Average
	Overall Accuracy	0.73
	Accuracy	0.91	0.76	0.79	0.82
	Precision	0.53	0.73	0.76	0.73
	Recall	0.56	0.56	0.88	0.73
	F1 score	0.55	0.64	0.82	0.73