[Enhancement] Mock LPIPS module in PPL metric (#1490)

* mock LPIPS module in PPL metric

* mock cpu test

tests/test_evaluation/test_metrics/test_ppl.py

@@ -1,17 +1,19 @@
 # Copyright (c) OpenMMLab. All rights reserved.
 import platform
-from unittest.mock import MagicMock
+from unittest.mock import patch
 
 import pytest
 import torch
-import torch.nn as nn
 from mmengine.runner import Runner
 
 from mmedit.datasets import BasicImageDataset
 from mmedit.datasets.transforms import PackEditInputs
 from mmedit.evaluation import PerceptualPathLength
 from mmedit.models import LSGAN, GenDataPreprocessor
 from mmedit.models.editors.stylegan2 import StyleGAN2Generator
+from mmedit.utils import register_all_modules
+
+register_all_modules()
 
 
 def process_fn(data_batch, predictions):
@@ -22,29 +24,17 @@ def process_fn(data_batch, predictions):
     return data_batch, _predictions
 
 
-class vgg_pytorch_classifier(nn.Module):
-
-    def __init__(self):
-        super().__init__()
-
-    def forward(self, x):
-        return torch.randn(x.shape[0], 4096)
+class LPIPS_mock:
 
+    def __init__(self, *args, **kwargs):
+        pass
 
-class vgg_mock(nn.Module):
+    def to(self, *args, **kwargs):
+        return self
 
-    def __init__(self, style):
-        super().__init__()
-        self.classifier = nn.Sequential(nn.Identity(), nn.Identity(),
-                                        nn.Identity(),
-                                        vgg_pytorch_classifier())
-        self.style = style
-
-    def forward(self, x, *args, **kwargs):
-        if self.style.upper() == 'STYLEGAN':
-            return torch.randn(x.shape[0], 4096)
-        else:  # torch
-            return torch.randn(x.shape[0], 7 * 7 * 512)
+    def __call__(self, x1, x2, *args, **kwargs):
+        num_batche = x1.shape[0]
+        return torch.rand(num_batche, 1, 1, 1)
 
 
 class TestPPL:
@@ -69,9 +59,7 @@ def setup_class(cls):
         generator = StyleGAN2Generator(64, 8)
         cls.module = LSGAN(generator, data_preprocessor=gan_data_preprocessor)
 
-        cls.mock_vgg_pytorch = MagicMock(
-            return_value=(vgg_mock('PyTorch'), 'False'))
-
+    @patch('lpips.LPIPS', LPIPS_mock)
     @pytest.mark.skipif(not torch.cuda.is_available(), reason='requires cuda')
     def test_ppl_cuda(self):
         ppl = PerceptualPathLength(
@@ -104,6 +92,7 @@ def test_ppl_cuda(self):
         ppl_res = ppl.compute_metrics(ppl.fake_results)
         assert ppl_res['ppl_score'] >= 0
 
+    @patch('lpips.LPIPS', LPIPS_mock)
     @pytest.mark.skipif(
         'win' in platform.system().lower() and 'cu' in torch.__version__,
         reason='skip on windows-cuda due to limited RAM.')