## Author Contributions

Conceptualization, S.S. and Y.A.-G.; Methodology, S.S., Y.A.-G., and G.G.-M.; Software, S.S.; Validation, S.S., Y.A.-G., and G.G.-M.; Formal Analysis, S.S., Y.A.-G. and G.G.-M.; Investigation, S.S., Y.A.-G., G.G.-M., A.R.-C., and J.M.M.-M.; Resources, S.S. and Y.A.-G.; Writing—Original Draft Preparation, S.S. and G.G.-M.; Writing—Review and Editing, G.G.-M., A.R.-C., and J.M.M.-M.; Supervision, Y.A.-G., A.R.-C., and J.M.M.-M.; Project Administration, Y.A.-G., A.R.-C., and J.M.M.-M.; Funding Acquisition, Y.A.-G., G.G.-M.

**Figure 1.**
Steps in the development of a complete algorithm for color segmentation of apples. In parentheses, a reference to the section/subsection where each step is described.

**Figure 1.**
Steps in the development of a complete algorithm for color segmentation of apples. In parentheses, a reference to the section/subsection where each step is described.

**Figure 2.**
Sample frames of videos taken under different lighting conditions. For each case, the weather condition and the measured light intensity (in lux) is indicated.

**Figure 2.**
Sample frames of videos taken under different lighting conditions. For each case, the weather condition and the measured light intensity (in lux) is indicated.

**Figure 3.**
Sample apple tree image and RGB channels. (**a**) Original RGB image. (**b**) R channel. (**c**) G channel. (**d**) B channel.

**Figure 3.**
Sample apple tree image and RGB channels. (**a**) Original RGB image. (**b**) R channel. (**c**) G channel. (**d**) B channel.

**Figure 4.**
Sample apple tree image under six different color spaces. (**a**) RGB color space. (**b**) HSV color space. (**c**) YCbCr color space. (**d**) CMY color space. (**e**) HSL color space. (**f**) L*u*v* color space. For visualization purposes, the three channels of each space are represented in the R, G, and B channels.

**Figure 4.**
Sample apple tree image under six different color spaces. (**a**) RGB color space. (**b**) HSV color space. (**c**) YCbCr color space. (**d**) CMY color space. (**e**) HSL color space. (**f**) L*u*v* color space. For visualization purposes, the three channels of each space are represented in the R, G, and B channels.

**Figure 5.**
Sample result of three texture features in an image of apples. (**a**) Original color image. (**b**) Image obtained by applying local range feature. (**c**) Local entropy feature. (**d**) Local standard deviation feature.

**Figure 5.**
Sample result of three texture features in an image of apples. (**a**) Original color image. (**b**) Image obtained by applying local range feature. (**c**) Local entropy feature. (**d**) Local standard deviation feature.

**Figure 6.**
Sample results of applying intensity segmentation on apple images. (**a**,**d**,**g**) Three color samples of the videos produced in 796, 1920, and 659 lux. (**b**,**e**,**h**) Results from the intensity transformation by selecting the R channel. (**c**,**f**,**i**) Segmentation images after applying the threshold.

**Figure 6.**
Sample results of applying intensity segmentation on apple images. (**a**,**d**,**g**) Three color samples of the videos produced in 796, 1920, and 659 lux. (**b**,**e**,**h**) Results from the intensity transformation by selecting the R channel. (**c**,**f**,**i**) Segmentation images after applying the threshold.

**Figure 7.**
Results of different orders of the color, texture and intensity steps on the segmentation algorithm. (**a**) Original color image of the apples. (**b**) Segmentation in texture, color and intensity order. (**c**) Segmentation in intensity, texture and color order. (**d**) Segmentation in color, texture, and intensity order.

**Figure 7.**
Results of different orders of the color, texture and intensity steps on the segmentation algorithm. (**a**) Original color image of the apples. (**b**) Segmentation in texture, color and intensity order. (**c**) Segmentation in intensity, texture and color order. (**d**) Segmentation in color, texture, and intensity order.

**Figure 8.**
Receiver operating characteristic (ROC) graph of the hybrid ANN-CA classification of frames for the 16 different classes, using color features.

**Figure 8.**
Receiver operating characteristic (ROC) graph of the hybrid ANN-CA classification of frames for the 16 different classes, using color features.

**Figure 9.**
Flowchart of the proposed apple segmentation algorithm.

**Figure 9.**
Flowchart of the proposed apple segmentation algorithm.

**Figure 10.**
Final segmentation results for five sample frames from different light intensities. (**a**,**c**,**e**,**g**,**i**) Original color images. (**b**,**d**,**f**,**h**,**j**) Corresponding segmented images, with background in black.

**Figure 10.**
Final segmentation results for five sample frames from different light intensities. (**a**,**c**,**e**,**g**,**i**) Original color images. (**b**,**d**,**f**,**h**,**j**) Corresponding segmented images, with background in black.

**Table 1.**
Characteristics of the videos captured under 16 different light conditions. The videos were obtained on different days. In the videos of evening and morning, the sky was clear.

**Table 1.**
Characteristics of the videos captured under 16 different light conditions. The videos were obtained on different days. In the videos of evening and morning, the sky was clear.

Case Number | Capture Date | Weather Condition | Light Intensity (lux) | Time of the Day | Video Length (min) | Total Number of Frames | Train/Test + Evaluation Frames |
---|

1 | 16 July 2017 | Cloudy | 1025 | 13:25 | 05:05 | 1830 | 1281/549 |

2 | 17 July 2017 | Sunny | 1863 | 11:10 | 10:04 | 3624 | 2537/1087 |

3 | 20 July 2017 | Sunny | 1958 | 14:30 | 01:03 | 378 | 265/113 |

4 | 23 July 2017 | Cloudy | 531 | 15:35 | 07:25 | 2670 | 1869/801 |

5 | 25 July 2017 | Sunny | 1694 | 16:36 | 10:17 | 3702 | 2592/1110 |

6 | 29 July 2017 | Evening | 796 | 18:05 | 12:09 | 4374 | 3062/1312 |

7 | 1 August 2017 | Sunny | 1415 | 10:15 | 11:05 | 3990 | 2793/1197 |

8 | 3 August 2017 | Sunny | 2150 | 13:15 | 03:14 | 1164 | 815/349 |

9 | 5 August 2017 | Sunny | 1920 | 12:00 | 12:25 | 4470 | 3129/1341 |

10 | 9 August 2017 | Cloudy | 827 | 14:10 | 07:23 | 2658 | 1861/797 |

11 | 10 August 2017 | Evening | 659 | 19:15 | 06:00 | 2160 | 1512/648 |

12 | 13 August 2017 | Morning | 316 | 07:25 | 18:04 | 6504 | 4553/1952 |

13 | 15 August 2017 | Very cloudy | 229 | 20:05 | 00:53 | 318 | 223/95 |

14 | 18 August 2017 | Sunny | 1369 | 09:05 | 06:34 | 2364 | 1655/709 |

15 | 20 August 2017 | Cloudy | 411 | 16:25 | 07:46 | 2796 | 1957/838 |

16 | 22 August 2017 | Cloudy | 384 | 17:15 | 02:32 | 912 | 639/274 |

**Table 2.**
Color features extracted for each pixel, related to vegetation indices. R_{n}, G_{n}, and B_{n} refer to normalized red, green, and blue, respectively.

**Table 2.**
Color features extracted for each pixel, related to vegetation indices. R_{n}, G_{n}, and B_{n} refer to normalized red, green, and blue, respectively.

Extracted Feature | Formula for Calculating the Feature |
---|

Normalized first component of RGB | R_{n} = R/(R + G + B) |

Normalized second component of RGB | G_{n} = G/(R + G + B) |

Normalized third component of RGB | B_{n} = B/(R + G + B) |

Gray channel | gray = 0.2898 × R + 0.5870 × G + 0.1140 × B |

Additional green [20] | EXG = 2 × G_{n} − R_{n} − B_{n} |

Additional red [21] | EXR = 1.4 × R_{n} − G_{n} |

Color index for extracted vegetation cover [3] | CIVE = 0.441 × R_{n} − 0.811 × G_{n} + 0.385 × B_{n} + 18.78 |

Subtraction between additional green and additional red [22] | EXGR = EXG − EXR |

Normalized difference index [23] | NDI = (G_{n} − R_{n})/(G_{n} + R_{n}) |

Green index minus blue [20] | GB = (G_{n} − B_{n}) |

Red-blue contrast [24] | RBI = (G_{n} − B_{n})/(G_{n} + B_{n}) |

Green-red index [24] | ERI = (R_{n} − G_{n}) × (R_{n} − B_{n}) |

Additional green index [24] | EGI = (G_{n} − R_{n}) × (G_{n} − B_{n}) |

Additional blue index [24] | EBI = (B_{n} − G_{n}) × (B_{n} − R_{n}) |

**Table 3.**
Parameters used in the multilayer perceptron neural network for the selection of the most effective color features.

**Table 3.**
Parameters used in the multilayer perceptron neural network for the selection of the most effective color features.

Parameter | Value |
---|

Number of layers | 2 |

Number of neurons | First layer: 8 |

Second layer: 12 |

Transfer functions | First layer: hyperbolic tangent sigmoid |

Second layer: hyperbolic tangent sigmoid |

Backpropagation network training function | Scaled conjugate gradient |

Backpropagation weight/bias learning function | Hebb with decay weight learning |

**Table 4.**
Values of the multilayer perceptron artificial neural network parameters adjusted by the cultural algorithm (ANN-CA method).

**Table 4.**
Values of the multilayer perceptron artificial neural network parameters adjusted by the cultural algorithm (ANN-CA method).

Parameter | Value |
---|

Number of hidden layers | 3 |

Number of neurons | First layer: 25 |

Second layer: 25 |

Third layer: 25 |

Transfer functions | First layer: hyperbolic tangent sigmoid |

Second layer: hyperbolic tangent sigmoid |

Third layer: hyperbolic tangent sigmoid |

Backpropagation network training function | Levenberg–Marquardt |

Backpropagation weight/bias learning function | LVQ1 weight learning |

**Table 5.**
Confusion matrix of the classification of 43,914 frames in the 16 classes corresponding to the different lighting conditions, using color features and the hybrid ANN-CA method. Rows: real classes. Columns: predicted classes. The percentage of color sharing of each class is defined as the percentage of frames of that class classified in a different class.

**Table 5.**
Confusion matrix of the classification of 43,914 frames in the 16 classes corresponding to the different lighting conditions, using color features and the hybrid ANN-CA method. Rows: real classes. Columns: predicted classes. The percentage of color sharing of each class is defined as the percentage of frames of that class classified in a different class.

| 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | Color Sharing by Class (%) | Total Color Sharing (%) |
---|

1 | 330 | 0 | 0 | 200 | 0 | 520 | 0 | 0 | 0 | 90 | 0 | 440 | 0 | 0 | 250 | 0 | 81.97 | 58.96 |

2 | 0 | 180 | 70 | 0 | 1330 | 0 | 90 | 104 | 1320 | 0 | 0 | 10 | 0 | 290 | 200 | 30 | 93.93 |

3 | 18 | 0 | 262 | 6 | 0 | 0 | 10 | 0 | 10 | 24 | 0 | 42 | 0 | 0 | 6 | 0 | 30.69 |

4 | 110 | 0 | 0 | 510 | 0 | 830 | 0 | 0 | 0 | 300 | 0 | 710 | 0 | 0 | 210 | 0 | 80.90 |

5 | 0 | 232 | 150 | 0 | 1710 | 0 | 50 | 30 | 990 | 0 | 0 | 0 | 0 | 520 | 0 | 20 | 53.81 |

6 | 80 | 0 | 0 | 160 | 0 | 2310 | 10 | 0 | 0 | 290 | 0 | 1300 | 0 | 0 | 224 | 0 | 47.19 |

7 | 0 | 50 | 20 | 30 | 160 | 130 | 2750 | 80 | 140 | 80 | 0 | 360 | 0 | 20 | 130 | 40 | 31.08 |

8 | 0 | 126 | 0 | 0 | 24 | 0 | 44 | 258 | 618 | 0 | 0 | 12 | 0 | 82 | 0 | 0 | 75.29 |

9 | 0 | 230 | 80 | 0 | 1270 | 0 | 170 | 380 | 1930 | 0 | 0 | 0 | 0 | 330 | 30 | 50 | 56.82 |

10 | 0 | 0 | 0 | 78 | 0 | 470 | 180 | 0 | 40 | 1030 | 0 | 740 | 0 | 0 | 120 | 0 | 61.25 |

11 | 0 | 0 | 10 | 20 | 0 | 120 | 1090 | 0 | 70 | 90 | 0 | 550 | 0 | 20 | 190 | 0 | 100 |

12 | 220 | 0 | 30 | 70 | 0 | 770 | 300 | 30 | 0 | 180 | 0 | 4330 | 0 | 0 | 574 | 0 | 33.27 |

13 | 30 | 0 | 12 | 0 | 0 | 41 | 8 | 0 | 0 | 65 | 0 | 132 | 6 | 0 | 24 | 0 | 98.11 |

14 | 0 | 30 | 10 | 0 | 710 | 0 | 20 | 80 | 580 | 0 | 0 | 24 | 0 | 890 | 0 | 20 | 62.35 |

15 | 30 | 0 | 10 | 90 | 0 | 760 | 20 | 0 | 0 | 90 | 0 | 860 | 0 | 0 | 936 | 0 | 66.52 |

16 | 0 | 10 | 0 | 0 | 40 | 0 | 110 | 30 | 72 | 0 | 0 | 30 | 10 | 30 | 90 | 490 | 44.05 |

**Table 6.**
Thresholds defined in RGB color space for the segmentation of background pixels.

**Table 6.**
Thresholds defined in RGB color space for the segmentation of background pixels.

Rule Number | R | | G | | B | | R-G | R-B | G-B |
---|

1 | [224,245] | and | [191,203] | and | [162,173] | and | - | - | [0,35] |

2 | [178,197] | and | [170,200] | and | [150,180] | and | - | [0,20] | - |

3 | [102,125] | and | [85,105] | and | [35,60] | and | [0,25] | - | - |

4 | [185,205] | and | [170,190] | and | [7,30] | and | [0,15] | - | - |

5 | [230,242] | and | [196,210] | and | [138,155] | and | [0,35] | - | - |

6 | [181,190] | and | [154,160] | and | [120,126] | and | [0,35] | - | - |

7 | [114,120] | and | [80,87] | and | [35,43] | and | [0,35] | - | - |

8 | [140,160] | and | [113,125] | and | [110,130] | and | - | - | [0,5] |

9 | [170,180] | and | [135,170] | and | [100,145] | and | - | - | [0,10] |

10 | [220,250] | and | [210,240] | and | [140,185] | and | | - | [0,10] |

11 | [208,230] | and | [195,210] | and | [125,135] | and | [0,20] | - | - |

12 | [150,165] | and | [140,155] | and | [37,50] | and | [0,15] | - | - |

13 | [200,225] | and | [180,200] | and | [155,180] | and | - | [0,20] | - |

14 | [215,230] | and | [155,170] | and | [130,150] | and | [0,30] | - | - |

15 | [185,195] | and | [170,185] | and | [120,130] | and | [0,20] | - | - |

16 | [142,167] | and | [120,139] | and | [67,97] | and | [0,30] | - | - |

17 | [170,185] | and | [155,170] | and | [13,30] | and | [0,20] | - | - |

18 | [200,222] | and | [175,202] | and | [110,125] | and | [0,35] | - | - |

19 | [198,225] | and | [186,210] | and | [50,76] | and | - | [0,20] | - |

20 | [178,182] | and | [164,167] | and | [126,133] | and | [0,18] | - | - |

21 | [98,116] | and | [70,98] | and | [50,83] | and | [0,30] | - | - |

22 | [129,140] | and | [110,120] | and | [109,120] | and | - | - | [0,10] |

23 | [239,255] | and | [230,255] | and | [55,170] | and | - | [0,25] | - |

24 | [180,255] | and | [180,255] | and | [180,255] | and | - | - | - |

25 | [160,200] | and | [149,180] | and | [30,60] | and | [0,20] | - | - |

26 | [200,220] | and | [190,210] | and | [120,145] | and | [0,15] | - | - |

27 | [125,140] | and | [117,130] | and | [25,42] | and | - | [0,15] | - |

28 | [170,190] | and | [140,155] | and | [130,140] | and | - | - | [0,20] |

29 | [178,185] | and | [125,140] | and | [105,125] | and | - | - | [0,25] |

30 | [195,220] | and | [185,210] | and | [135,155] | and | [0,15] | - | - |

31 | [220,235] | and | [195,215] | and | [150,180] | and | [0,30] | - | - |

32 | [92,102] | and | [82,94] | and | [20,35] | and | [0,15] | - | - |

33 | [100,110] | and | [75,90] | and | [68,82] | and | - | - | [0,15] |

34 | [220,230] | and | [193,207] | and | [125,145] | and | [0,30] | - | - |

35 | [212,230] | and | [202,220] | and | [20,38] | and | [0,15] | - | - |

36 | [105,108] | and | [74,77] | and | [64,66] | and | - | - | [0,15] |

37 | [96,100] | and | [74,78] | and | [66,70] | and | [0,10] | - | - |

38 | [95,127] | and | [91,110] | and | [85,90] | and | [0,35] | - | - |

39 | [150,155] | and | [136,140] | and | [119,122] | and | - | - | [0,20] |

40 | [146,166] | and | [133,150] | and | [49,54] | and | [0,20] | - | - |

41 | [97,110] | and | [84,95] | and | [30,45] | and | [0,15] | - | - |

42 | [120,135] | and | [100,115] | and | [55,68] | and | [0,25] | - | - |

43 | [120,155] | and | [95,120] | and | [80,100] | and | - | - | [0,25] |

44 | [195,210] | and | [160,185] | and | [125,150] | and | [50,70] | - | - |

45 | [110,115] | and | [75,78] | and | [57,78] | and | - | - | [0,20] |

46 | [115,130] | and | [100,115] | and | [49,55] | and | [0,20] | - | - |

47 | [132,142] | and | [78,85] | and | [52,58] | and | - | - | [0,30] |

48 | [92,130] | and | [63,95] | and | [12,35] | and | [0,35] | - | - |

49 | [220,225] | and | [226,242] | and | [102,110] | and | - | [0,20] | - |

50 | [127,142] | and | [96,110] | and | [68,77] | and | [0,35] | - | - |

51 | [120,140] | and | [85,110] | and | [50,67] | and | - | [0,40] | - |

52 | [100,160] | and | [100,160] | and | [10,50] | and | [0,10] | - | - |

53 | [189,200] | and | [171,191] | and | [60,9] | and | [0,15] | - | - |

54 | [105,120] | and | [85,100] | and | [48,60] | and | [0,25] | - | - |

55 | [225,235] | and | [170,180] | and | [135,145] | and | - | - | [0,35] |

56 | [215,220] | and | [195,205] | and | [150,160] | and | [0,25] | - | - |

57 | [190,202] | and | [179,190] | and | [48,53] | and | [0,20] | - | - |

58 | [109,122] | and | [68,90] | and | [42,58] | and | - | - | [0,35] |

59 | [109,120] | and | [65,80] | and | [69,75] | and | - | [0,35] | - |

60 | [129,140] | and | [89,95] | and | [64,75] | and | - | - | [0,27] |

61 | [191,195] | and | [170,178] | and | [128,132] | and | [0,25] | - | - |

62 | [155,170] | and | [132,152] | and | [100,125] | and | [0,35] | - | - |

63 | [95,105] | and | [57,72] | and | [27,42] | and | - | - | [0,35] |

64 | [110,135] | and | [90,120] | and | [85,110] | and | - | - | [0,15] |

65 | [250,255] | and | [182,188] | and | [146,149] | and | - | - | [0,40] |

66 | [215,220] | and | [183,190] | and | [135,140] | and | [0,35] | - | - |

67 | [113,122] | and | [87,93] | and | [81,87] | and | - | - | [0,10] |

68 | [227,242] | and | [211,225] | and | [72,82] | and | [0,20] | - | - |

69 | [164,172] | and | [150,159] | and | [95,105] | and | [0,20] | - | - |

70 | [139,145] | and | [111,120] | and | [70,81] | and | [0,35] | - | - |

71 | [142,152] | and | [118,127] | and | [90,107] | and | - | - | [0,30] |

72 | - | - | - | - | - | - | [0,10] | - | - |

73 | [122,131] | and | [105,110] | and | [56,60] | and | [0,22] | - | - |

74 | [230,245] | and | [165,175] | and | [120,130] | and | - | - | [35,50] |

75 | [225,235] | and | [170,180] | and | [135,145] | and | - | [0,40] | - |

76 | [90,110] | and | [55,80] | and | [40,62] | and | - | - | [0,20] |

77 | [195,207] | and | [179,182] | and | [25,33] | and | [0,27] | - | - |

78 | [244,248] | and | [232,236] | and | [84,88] | and | [0,15] | - | - |

79 | [181,189] | and | [169,178] | and | [113,121] | and | [0,15] | - | - |

80 | [190,210] | and | [180,205] | and | [110,125] | and | [0,20] | - | - |

81 | [190,210] | and | [160,180] | and | [140,170] | and | - | - | [0,20] |

82 | [212,220] | and | [193,205] | and | [125,148] | and | [0,25] | - | - |

83 | [98,108] | and | [82,90] | and | [22,38] | and | [0,25] | - | - |

84 | [202,227] | and | [185,215] | and | [35,40] | and | [0,20] | - | - |

85 | [215,237] | and | [215,227] | and | [45,65] | and | - | [0,15] | - |

86 | [170,205] | and | [165,185] | and | [87,100] | and | [0,20] | - | - |

87 | [125,160] | and | [110,145] | and | [55,90] | and | [0,20] | - | - |

88 | [222,230] | and | [161,169] | and | [130,137] | and | - | - | [0,35] |

89 | [155,185] | and | [130,165] | and | [110,150] | and | - | - | [0,20] |

90 | [195,215] | and | [165,195] | and | [120,140] | and | [0,35] | - | - |

91 | [120,128] | and | [110,118] | and | [40,55] | and | [0,15] | - | - |

92 | [95,115] | and | [49,70] | and | [25,50] | and | - | - | [0,25] |

**Table 7.**
Confusion matrix and accuracy of the proposed apple segmentation algorithm in the test set.

**Table 7.**
Confusion matrix and accuracy of the proposed apple segmentation algorithm in the test set.

Predicted/Real Class | Apple | Background | All Data | Classification Error by Class (%) | Classification Accuracy (%) |
---|

Apple | 91,406 | 798 | 92,204 | 0.865 | 99.12 |

Background | 1052 | 117,496 | 118,548 | 0.887 |

**Table 8.**
Confusion matrix and accuracy of the segmentation algorithm using neural networks on the test set.

**Table 8.**
Confusion matrix and accuracy of the segmentation algorithm using neural networks on the test set.

Predicted/Real Class | Apple | Background | All Data | Classification Error by Class (%) | Classification Accuracy (%) |
---|

Apple | 87,090 | 5114 | 92,204 | 5.55 | 95.23 |

Background | 4932 | 113,616 | 118,548 | 4.16 |

**Table 9.**
Confusion matrix and accuracy of the segmentation algorithm using color histogram models on the testing set.

**Table 9.**
Confusion matrix and accuracy of the segmentation algorithm using color histogram models on the testing set.

Predicted/Real Class | Apple | Background | All Data | Classification Error by Class (%) | Classification Accuracy (%) |
---|

Apple | 88,103 | 4101 | 92,204 | 4.45 | 96.80 |

Background | 2645 | 115,903 | 118,548 | 2.23 |

**Table 10.**
Measures of sensitivity (Sens.), specificity (Spec.) and accuracy (Accur.) of the proposed segmentation algorithm and the two methods used for comparison: neural networks and color histogram models.

**Table 10.**
Measures of sensitivity (Sens.), specificity (Spec.) and accuracy (Accur.) of the proposed segmentation algorithm and the two methods used for comparison: neural networks and color histogram models.

| Proposed Method | Neural Networks | Color Histograms |
---|

Class | Sens. (%) | Spec. (%) | Accur. (%) | Sens. (%) | Spec. (%) | Accur. (%) | Sens. (%) | Spec. (%) | Accur. (%) |
---|

Apple | 99.13 | 98.86 | 99.12 | 94.45 | 94.64 | 95.23 | 95.55 | 97.09 | 96.80 |

Background | 99.11 | 99.33 | 95.84 | 95.69 | 97.77 | 96.58 |

**Table 11.**
Comparison of the segmentation accuracy of the proposed algorithm with respect to other recent research works.

**Table 11.**
Comparison of the segmentation accuracy of the proposed algorithm with respect to other recent research works.

Method | Number of Test Samples | Accuracy Rate (%) |
---|

Proposed method | 210,752 | 99.12 |

ANN method | 210,752 | 95.23 |

Color histograms method | 210,752 | 96.8 |

Tang et al. [30] | 100 | 92.5 |

Aquino et al. [31] | 152 | 95.72 |