[Fix][New_config] Fix demo bug (#1647)

* Fix demo

* Update implement

---------

Co-authored-by: mzr1996 <mzr1996@163.com>

mmpretrain/apis/feature_extractor.py

@@ -83,9 +83,11 @@ class FeatureExtractor(BaseInferencer):
 
     def _init_pipeline(self, cfg: Config) -> Callable:
         test_pipeline_cfg = cfg.test_dataloader.dataset.pipeline
-        if test_pipeline_cfg[0]['type'] == 'LoadImageFromFile':
+        from mmpretrain.datasets import remove_transform
-            # Image loading is finished in `self.preprocess`.
+
-            test_pipeline_cfg = test_pipeline_cfg[1:]
+        # Image loading is finished in `self.preprocess`.
+        test_pipeline_cfg = remove_transform(test_pipeline_cfg,
+                                             'LoadImageFromFile')
         test_pipeline = Compose(
             [TRANSFORMS.build(t) for t in test_pipeline_cfg])
         return test_pipeline

mmpretrain/apis/image_caption.py

@@ -70,9 +70,11 @@ class ImageCaptionInferencer(BaseInferencer):
 
     def _init_pipeline(self, cfg: Config) -> Callable:
         test_pipeline_cfg = cfg.test_dataloader.dataset.pipeline
-        if test_pipeline_cfg[0]['type'] == 'LoadImageFromFile':
+        from mmpretrain.datasets import remove_transform
-            # Image loading is finished in `self.preprocess`.
+
-            test_pipeline_cfg = test_pipeline_cfg[1:]
+        # Image loading is finished in `self.preprocess`.
+        test_pipeline_cfg = remove_transform(test_pipeline_cfg,
+                                             'LoadImageFromFile')
         test_pipeline = Compose(
             [TRANSFORMS.build(t) for t in test_pipeline_cfg])
         return test_pipeline

mmpretrain/apis/image_classification.py

@@ -110,9 +110,11 @@ class ImageClassificationInferencer(BaseInferencer):
 
     def _init_pipeline(self, cfg: Config) -> Callable:
         test_pipeline_cfg = cfg.test_dataloader.dataset.pipeline
-        if test_pipeline_cfg[0]['type'] == 'LoadImageFromFile':
+        from mmpretrain.datasets import remove_transform
-            # Image loading is finished in `self.preprocess`.
+
-            test_pipeline_cfg = test_pipeline_cfg[1:]
+        # Image loading is finished in `self.preprocess`.
+        test_pipeline_cfg = remove_transform(test_pipeline_cfg,
+                                             'LoadImageFromFile')
         test_pipeline = Compose(
             [TRANSFORMS.build(t) for t in test_pipeline_cfg])
         return test_pipeline

mmpretrain/apis/image_retrieval.py

@@ -172,9 +172,11 @@ class ImageRetrievalInferencer(BaseInferencer):
 
     def _init_pipeline(self, cfg: Config) -> Callable:
         test_pipeline_cfg = cfg.test_dataloader.dataset.pipeline
-        if test_pipeline_cfg[0]['type'] == 'LoadImageFromFile':
+        from mmpretrain.datasets import remove_transform
-            # Image loading is finished in `self.preprocess`.
+
-            test_pipeline_cfg = test_pipeline_cfg[1:]
+        # Image loading is finished in `self.preprocess`.
+        test_pipeline_cfg = remove_transform(test_pipeline_cfg,
+                                             'LoadImageFromFile')
         test_pipeline = Compose(
             [TRANSFORMS.build(t) for t in test_pipeline_cfg])
         return test_pipeline

mmpretrain/apis/visual_grounding.py

@@ -86,9 +86,11 @@ class VisualGroundingInferencer(BaseInferencer):
 
     def _init_pipeline(self, cfg: Config) -> Callable:
         test_pipeline_cfg = cfg.test_dataloader.dataset.pipeline
-        if test_pipeline_cfg[0]['type'] == 'LoadImageFromFile':
+        from mmpretrain.datasets import remove_transform
-            # Image loading is finished in `self.preprocess`.
+
-            test_pipeline_cfg = test_pipeline_cfg[1:]
+        # Image loading is finished in `self.preprocess`.
+        test_pipeline_cfg = remove_transform(test_pipeline_cfg,
+                                             'LoadImageFromFile')
         test_pipeline = Compose(
             [TRANSFORMS.build(t) for t in test_pipeline_cfg])
         return test_pipeline

mmpretrain/apis/visual_question_answering.py

@@ -88,9 +88,11 @@ class VisualQuestionAnsweringInferencer(BaseInferencer):
 
     def _init_pipeline(self, cfg: Config) -> Callable:
         test_pipeline_cfg = cfg.test_dataloader.dataset.pipeline
-        if test_pipeline_cfg[0]['type'] == 'LoadImageFromFile':
+        from mmpretrain.datasets import remove_transform
-            # Image loading is finished in `self.preprocess`.
+
-            test_pipeline_cfg = test_pipeline_cfg[1:]
+        # Image loading is finished in `self.preprocess`.
+        test_pipeline_cfg = remove_transform(test_pipeline_cfg,
+                                             'LoadImageFromFile')
         test_pipeline = Compose(
             [TRANSFORMS.build(t) for t in test_pipeline_cfg])
         return test_pipeline

mmpretrain/datasets/transforms/__init__.py

@@ -16,6 +16,7 @@ from .processing import (Albumentations, BEiTMaskGenerator, CleanCaption,
                          RandomErasing, RandomResizedCrop,
                          RandomResizedCropAndInterpolationWithTwoPic,
                          RandomTranslatePad, ResizeEdge, SimMIMMaskGenerator)
+from .utils import get_transform_idx, remove_transform
 from .wrappers import ApplyToList, MultiView
 
 for t in (CenterCrop, LoadImageFromFile, Normalize, RandomFlip,
@@ -34,5 +35,6 @@ __all__ = [
     'SimMIMMaskGenerator', 'CenterCrop', 'LoadImageFromFile', 'Normalize',
     'RandomFlip', 'RandomGrayscale', 'RandomResize', 'Resize', 'MultiView',
     'ApplyToList', 'CleanCaption', 'RandomTranslatePad',
-    'RandomResizedCropAndInterpolationWithTwoPic'
+    'RandomResizedCropAndInterpolationWithTwoPic', 'get_transform_idx',
+    'remove_transform'
 ]

mmpretrain/datasets/transforms/utils.py

@@ -0,0 +1,53 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import copy
+from typing import List, Union
+
+from mmcv.transforms import BaseTransform
+
+PIPELINE_TYPE = List[Union[dict, BaseTransform]]
+
+
+def get_transform_idx(pipeline: PIPELINE_TYPE, target: str) -> int:
+    """Returns the index of the transform in a pipeline.
+
+    Args:
+        pipeline (List[dict] | List[BaseTransform]): The transforms list.
+        target (str): The target transform class name.
+
+    Returns:
+        int: The transform index. Returns -1 if not found.
+    """
+    for i, transform in enumerate(pipeline):
+        if isinstance(transform, dict):
+            if isinstance(transform['type'], type):
+                if transform['type'].__name__ == target:
+                    return i
+            else:
+                if transform['type'] == target:
+                    return i
+        else:
+            if transform.__class__.__name__ == target:
+                return i
+
+    return -1
+
+
+def remove_transform(pipeline: PIPELINE_TYPE, target: str, inplace=False):
+    """Remove the target transform type from the pipeline.
+
+    Args:
+        pipeline (List[dict] | List[BaseTransform]): The transforms list.
+        target (str): The target transform class name.
+        inplace (bool): Whether to modify the pipeline inplace.
+
+    Returns:
+        The modified transform.
+    """
+    idx = get_transform_idx(pipeline, target)
+    if not inplace:
+        pipeline = copy.deepcopy(pipeline)
+    while idx >= 0:
+        pipeline.pop(idx)
+        idx = get_transform_idx(pipeline, target)
+
+    return pipeline

tools/misc/verify_dataset.py

@@ -52,9 +52,12 @@ class DatasetValidator():
     def __init__(self, dataset_cfg, log_file_path):
         super(DatasetValidator, self).__init__()
         # keep only LoadImageFromFile pipeline
-        assert dataset_cfg.pipeline[0]['type'] == 'LoadImageFromFile', (
+        from mmpretrain.datasets import get_transform_idx
-            'This tool is only for datasets needs to load image from files.')
+
-        self.pipeline = TRANSFORMS.build(dataset_cfg.pipeline[0])
+        load_idx = get_transform_idx(dataset_cfg.pipeline, 'LoadImageFromFile')
+        assert load_idx >= 0, \
+            'This tool is only for datasets needs to load image from files.'
+        self.pipeline = TRANSFORMS.build(dataset_cfg.pipeline[load_idx])
         dataset_cfg.pipeline = []
         dataset = build_dataset(dataset_cfg)