Skip to main content
. 2024 May 3;11(5):454. doi: 10.3390/bioengineering11050454

Figure 2.

Figure 2

Illustration of network architectures for BM detection and segmentation. (a) SSD network. This example shows the dimensions of YOLO. The last layer outputs predicted bounding boxes and class probability with data dimensions of 7 × 7 × 30. (b) DeepMedic, an FCN type network. It combines a normal resolution and a low-resolution branch with output corresponding to the receptive fields. (c) U-Net. The encoder branch and decoder branch are connected at the same depth. Here, we illustrate U-Net with a depth of two.