Table 10 Performance of the GRU-based classifier in the signer-independent mode after more data are added.

From: A deep learning framework for Ethiopian sign language recognition using skeleton-based representation

Label

Precision

Recall

F1-Score

Support

ሀዘን

0.91

0.97

0.94

73

ህመም

0.50

0.67

0.57

70

ለምን?

0.48

0.97

0.64

76

ልምምድ

0.72

0.65

0.68

74

መራራ

0.92

1.00

0.96

72

መናደድ

0.86

0.55

0.67

67

መጀመር

1.00

0.52

0.69

71

መገረም

0.82

1.00

0.90

72

መጥፎ

0.79

0.66

0.72

73

መጨረስ

0.89

0.81

0.85

73

መጨነቅ

0.79

0.59

0.67

70

ማሸነፍ

0.65

0.67

0.66

73

ምስጋና

0.77

0.48

0.59

71

ቀላል

0.51

0.97

0.67

74

ተነስ!

0.80

0.64

0.71

75

አስቸጋሪ

0.93

0.85

0.89

75

እህት

0.87

0.52

0.65

66

ወንድም

0.81

0.63

0.71

73

ይቅርታ

0.80

1.00

0.89

69

ጎበዝ

0.76

0.52

0.62

73

Accuracy

  

0.74

1440

Macro Avg

0.78

0.73

0.73

1440

Weighted Avg

0.78

0.74

0.73

1440