Inception-LSTM Human Motion Recognition with Channel Attention Mechanism

Table 3

Experimental results of different models on self-built dataset.

Algorithm model	Accuracy	Precision	Recall	F1 value	Model size
Classics CNN	95.71%	96.12%	95.71%	95.91%	2.53 M
LSTM	95.02%	95.98%	95.02%	95.50%	0.34 M
Neural network without ECA	96.77%	96.85%	96.77%	96.81%	1.21 M
Original ECA	98.44%	98.54%	98.44%	98.49%	1.24 M
The proposed model	98.81%	98.81%	98.81%	98.81%	1.24 M