Skip to main content
Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

< Back to Article

Table 1.

The whole framework of Dense MobileNetV3.

More »

Table 1 Expand

Fig 1.

Depthwise separable convolution.

(a) Depthwise convolution. (b) Pointwise convolution.

More »

Fig 1 Expand

Fig 2.

Squeeze and Excitation (SE) module.

After excitation, different colors mean channels get different weights.

More »

Fig 2 Expand

Fig 3.

Improved bottleneck block.

The dark green area on the right is the detailed process of Dense Block.

More »

Fig 3 Expand

Fig 4.

Pruning procedure.

(a)Initial network. (b) Compact network. The compact network after pruning is fine-tuned to reach similar (or even better) accuracy than when trained normally.

More »

Fig 4 Expand

Table 2.

Specific division of the database.

More »

Table 2 Expand

Fig 5.

Some cases of CXR from the COVID-19 Radiography Database.

(A) COVID-19 sample, (B) Pneumonia sample, and (C) Normal sample.

More »

Fig 5 Expand

Table 3.

Hyper-parameter used during training.

More »

Table 3 Expand

Fig 6.

The fluctuation of accuracy and loss over epochs during training and validation.

More »

Fig 6 Expand

Fig 7.

Confusion matrices distribution.

More »

Fig 7 Expand

Table 4.

Ablation study metrics.

More »

Table 4 Expand

Table 5.

Comparison of the effect after using various pruning methods.

More »

Table 5 Expand

Table 6.

Comparison of parameters before and after pruning.

More »

Table 6 Expand

Table 7.

Dataset of five-fold cross-validation.

More »

Table 7 Expand

Table 8.

Results of five-fold cross-validation.

More »

Table 8 Expand

Table 9.

Evaluate accuracy and parameters by comparing the relevant model.

More »

Table 9 Expand

Fig 8.

F1-score comparison.

More »

Fig 8 Expand

Fig 9.

Visualization of chest X-ray images using Grad-CAM on Dense MobileNetV3 model.

More »

Fig 9 Expand