Skip to main content

Table 3 Performance comparison of global projection methods in ancestry estimation

From: ANINet: a deep neural network for skull ancestry estimation

Convolutional network Accuracy (%) Precision (%) Recall (%) F1-score (%) Specificity (%) Para size (MB)
AlexNet 95.04 92.26 96.27 94.80 92.01 233
Vgg-16 60.78 64.94 57.68 61.17 62.31 528
GoogLenet 96.08 97.31 94.41 95.96 93.69 51
ResNet-50 98.03 98.06 98.38 98.50 97.98 101
DenseNet-121 98.09 98.15 98.43 98.70 98.85 46
SqueezeNet 95.91 96.89 96.47 96.80 93.00 5
ANINet (SE-Wide) 98.21 98.49 98.51 98.77 99.09 45
  1. The significance of bold means that in a certain aspect (performance, model size), this model is the best among these models