Table 1 Comparison of generator networks using the number of parameters, inference time, and number of floating-point operations (FLOPs).

From: Robust resolution improvement of 3D UTE-MR angiogram of normal vasculatures using super-resolution convolutional neural network

Criteria

Networks

SR-ResNet

MRDG64

LSRDG

Number of parameters

 ~ 0.53 M

 ~ 6.95 M

 ~ 1.42 M

Inference time

 ~ 6.5 s

 ~ 31 s

 ~ 18 s

FLOPs

 ~ 2.8 TFLOPs

 ~ 16.2 TFLOPs

 ~ 5.5 TFLOPs

  1. Inference time and FLOPs refer to the time and number of operations to convert an image of size 1283 into 2563 on a central processing unit with 512 gigabytes of random access memory, while M and T refer to 106 and 1012, respectively.
Back to article page