Table 3 RMSE outcome of the HDLIP-SHAR approach with existing models under various mural images.

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

Murals (Image_ID)

RMSE

AOT Method

EC Model

LaMa Method

LID-MIR

HDLIP-SHAR

ID_001848

0.2197

0.1803

0.2606

0.1694

0.1249

ID_002224

0.1735

0.2764

0.1581

0.2330

0.1044

ID_002796

0.2264

0.2520

0.2829

0.2803

0.1939

ID_003162

0.2067

0.2371

0.2879

0.2655

0.1676

ID_003574

0.2268

0.2941

0.2943

0.2959

0.2166

ID_004277

0.1469

0.2314

0.2454

0.2406

0.1058

Back to article page