Human Activity Recognition Using a CNN with an Enhanced Convolutional Block Attention Module | Wuhan University Journal of Natural Sciences

Open Access

Table 1

Performance comparison of different models on the UT-HAR and NTU-Fi_HAR datasets (unit:%)

Method	UT-HAR		NTU-Fi_HAR
Method	Acc.	F₁	Acc.	F₁
Modified LeNet-5	94.20	94.17	97.73	97.72
ResNet-18	98.60	98.59	87.50	87.30
LSTM	87.00	86.77	95.83	95.79
ABLSTM	94.40	94.39	96.59	96.54
GAF-CNN	68.00	66.83	99.62	99.62
CNN-ECBAM	$99 . 20^{*}$	$99 . 20^{*}$	$100^{* *}$	$100^{* *}$

Note: Statistical significance vs. strongest baseline is indicated by * (p<0.05), ** (p<0.01) using paired t-test over 5 independent runs.

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.