Skip to main content
. 2022 Aug 28;22(17):6481. doi: 10.3390/s22176481

Table 5.

Performance of the pretrained networks DenseNet-121 and VGG-16 considering the test sets for the classification scenario of 730 images (without lesions and with lesions between 0.5 and 1.9 mm) from the UFPE database. The best values are presented in bold.

Methods Metrics
Accuracy F1-score Specificity Precision Recall
DenseNet-121 0.6667 0.4494 0.8571 0.6060 0.3571
VGG-16 0.6599 0.4318 0.8041 0.5000 0.3800