Investigation on Recognition Performance of Harvesting Robot Using Regions of Interest Histogram of Oriented Gradients Feature Based on Improved Fuzzy Least Square Support Vector Machine

Table 2

Comparisons of experimental identification results between the FLS-SVM, the SVM, and the faster R-CNN.

Identification methods and identification results		Identification images
		Learning samples	Isolated fruit test sample	Overlap fruit sample	Environmental background samples
Image total		4000	2000	1000	3000
SVM identification results	Correct amount	3945	1862	856	2850
	Number of errors for environmental background	45	110	110	30
	Number of errors for fruit background	10	28	34	120
	Accuracy	98.60%	93.10%	85.60%	95.00%
FLS-SVM identification results	Correct amount	3982	1920	894	2910
	Number of errors for environmental background	16	80	106	0
	Number of errors for fruit background	2	0	0	90
	Accuracy	99.50%	96.00%	89.40%	97.00%
Faster R-CNN identification results	Correct amount	3979	1916	890	2906
	Number of errors for environmental background	17	82	108	2
	Number of errors for fruit background	4	2	2	92
	Accuracy	99.48%	95.80%	89.00%	96.87%