Table 12 Comparison of computational efficiency, including FLOPs, GPU usage, and inference time.

From: Automated image inpainting for historical artifact restoration using hybridisation of transfer learning with deep generative models

Methods

FLOPs (G)

GPU (M)

Inference Time (sec)

CNN

22.8

4533

2.79

ResNet

24.7

2295

5.01

ViT

32.2

2462

5.15

PSMNet

26.7

3363

3.28

HDLIP-SHAR

10.11

755

1.08

Back to article page