Table 1.

Total number and relative percentage of data samples within the training, validation and test sets.

		Training		Validation		Testing		Total
Total		2,431	69.46%	705	20.14%	364	10.40%	3,500
Label	Incomplete	311	66.17%	101	21.49%	58	12.34%	470
	Twisted	274	71.54%	76	19.84%	33	8.62%	383
	Too long	345	66.99%	102	19.81%	68	13.20%	515
	Too loose	40	60.61%	17	25.76%	9	13.63%	66
	Below GE junction	261	65.74%	83	20.91%	53	13.35%	397
	Correct	1,200	71.90%	326	19.53%	143	8.57%	1,669

Images with more than 1 VQI are counted separately; hence, the total exceeds the 3,138 samples.