Skip to main content

Table 2 Performance comparison of local 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 90.83 89.38 93.17 91.97 87.46 233
Vgg16 62.17 69.70 57.76 63.23 67.89 528
GoogLenet 95.92 96.11 96.58 96.65 95.01 51
ResNet-50 96.33 96.74 98.09 97.15 95.09 101
DenseNet-121 97.97 97.98 98.15 98.19 98.55 46
SqueezeNet 92.91 93.76 93.58 93.70 90.88 5
ANINet (SE-Wide) 98.04 98.76 98.44 98.27 99.01 45
  1. The significance of bold means that in a certain aspect (performance, model size), this model is the best among these models