Skip to main content
. 2022 Oct 18;19(20):13491. doi: 10.3390/ijerph192013491

Figure 3.

Figure 3

Model training architectures (system units) with the ResNet-34, EfficientNet B4, and ECA-Net50 backbones: (a) Channel A: with pose estimator; (b) Channel B: without pose estimator.