Skip to main content

Table 5 Low-level action results using base classifiers and LSTM network

From: CAPHAR: context-aware personalized human activity recognition using associative learning in smart environments

Action

ECOC (Tree)

ECOC (QDA)

ECOC (AB)

LSTM

Standing

91.06%

91.18%

93.60%

92.67%

Sitting

94.45%

93.56%

95.97%

93.84%

Walking

94.49%

95.02%

93.50%

89.45%

Climbing (up)

82.93%

83.74%

87.12%

86.97%

Climbing (down)

91.80%

90.99%

92.45%

92.07%

Lying

96.65%

94.38%

97.67%

97.34%

Running

84.57%

86.84%

88.61%

84.16%

Average

90.85%

90.81%

92.70%

90.93%

  1. ECOC, Error Correcting Output Codes; QDA, Quadratic Discriminant Analysis; AB, Adaptive Boosting; LSTM, Long Short-Term Memory
Back to article page