use a single for loop (huggingface#33148)

* use a single for loop * oups * fixup * fix typo
Cyrilvallez · Aug 29, 2024 · c409cd8 · c409cd8
1 parent 5129671
commit c409cd8
Show file tree

Hide file tree

Showing 10 changed files with 168 additions and 218 deletions.
diff --git a/src/transformers/models/bit/image_processing_bit.py b/src/transformers/models/bit/image_processing_bit.py
@@ -294,31 +294,27 @@ def preprocess(
             # We assume that all images have the same channel dimension format.
             input_data_format = infer_channel_dimension_format(images[0])
 
-        if do_resize:
-            images = [
-                self.resize(image=image, size=size, resample=resample, input_data_format=input_data_format)
-                for image in images
-            ]
-
-        if do_center_crop:
-            images = [
-                self.center_crop(image=image, size=crop_size, input_data_format=input_data_format) for image in images
-            ]
-
-        if do_rescale:
-            images = [
-                self.rescale(image=image, scale=rescale_factor, input_data_format=input_data_format)
-                for image in images
-            ]
-
-        if do_normalize:
-            images = [
-                self.normalize(image=image, mean=image_mean, std=image_std, input_data_format=input_data_format)
-                for image in images
-            ]
+        all_images = []
+        for image in images:
+            if do_resize:
+                image = self.resize(image=image, size=size, resample=resample, input_data_format=input_data_format)
+
+            if do_center_crop:
+                image = self.center_crop(image=image, size=crop_size, input_data_format=input_data_format)
+
+            if do_rescale:
+                image = self.rescale(image=image, scale=rescale_factor, input_data_format=input_data_format)
+
+            if do_normalize:
+                image = self.normalize(
+                    image=image, mean=image_mean, std=image_std, input_data_format=input_data_format
+                )
+
+            all_images.append(image)
 
         images = [
-            to_channel_dimension_format(image, data_format, input_channel_dim=input_data_format) for image in images
+            to_channel_dimension_format(image, data_format, input_channel_dim=input_data_format)
+            for image in all_images
         ]
 
         data = {"pixel_values": images}

diff --git a/src/transformers/models/chameleon/image_processing_chameleon.py b/src/transformers/models/chameleon/image_processing_chameleon.py
@@ -311,32 +311,26 @@ def preprocess(
         if input_data_format is None:
             # We assume that all images have the same channel dimension format.
             input_data_format = infer_channel_dimension_format(images[0])
+        all_images = []
+        for image in images:
+            if do_resize:
+                image = self.resize(image=image, size=size, resample=resample, input_data_format=input_data_format)
 
-        if do_resize:
-            images = [
-                self.resize(image=image, size=size, resample=resample, input_data_format=input_data_format)
-                for image in images
-            ]
-
-        if do_center_crop:
-            images = [
-                self.center_crop(image=image, size=crop_size, input_data_format=input_data_format) for image in images
-            ]
-
-        if do_rescale:
-            images = [
-                self.rescale(image=image, scale=rescale_factor, input_data_format=input_data_format)
-                for image in images
-            ]
-
-        if do_normalize:
-            images = [
-                self.normalize(image=image, mean=image_mean, std=image_std, input_data_format=input_data_format)
-                for image in images
-            ]
+            if do_center_crop:
+                image = self.center_crop(image=image, size=crop_size, input_data_format=input_data_format)
 
+            if do_rescale:
+                image = self.rescale(image=image, scale=rescale_factor, input_data_format=input_data_format)
+
+            if do_normalize:
+                image = self.normalize(
+                    image=image, mean=image_mean, std=image_std, input_data_format=input_data_format
+                )
+
+            all_images.append(image)
         images = [
-            to_channel_dimension_format(image, data_format, input_channel_dim=input_data_format) for image in images
+            to_channel_dimension_format(image, data_format, input_channel_dim=input_data_format)
+            for image in all_images
         ]
 
         data = {"pixel_values": images}

diff --git a/src/transformers/models/chinese_clip/image_processing_chinese_clip.py b/src/transformers/models/chinese_clip/image_processing_chinese_clip.py
@@ -280,31 +280,26 @@ def preprocess(
             # We assume that all images have the same channel dimension format.
             input_data_format = infer_channel_dimension_format(images[0])
 
-        if do_resize:
-            images = [
-                self.resize(image=image, size=size, resample=resample, input_data_format=input_data_format)
-                for image in images
-            ]
+        all_images = []
+        for image in images:
+            if do_resize:
+                image = self.resize(image=image, size=size, resample=resample, input_data_format=input_data_format)
 
-        if do_center_crop:
-            images = [
-                self.center_crop(image=image, size=crop_size, input_data_format=input_data_format) for image in images
-            ]
+            if do_center_crop:
+                image = self.center_crop(image=image, size=crop_size, input_data_format=input_data_format)
 
-        if do_rescale:
-            images = [
-                self.rescale(image=image, scale=rescale_factor, input_data_format=input_data_format)
-                for image in images
-            ]
+            if do_rescale:
+                image = self.rescale(image=image, scale=rescale_factor, input_data_format=input_data_format)
 
-        if do_normalize:
-            images = [
-                self.normalize(image=image, mean=image_mean, std=image_std, input_data_format=input_data_format)
-                for image in images
-            ]
+            if do_normalize:
+                image = self.normalize(
+                    image=image, mean=image_mean, std=image_std, input_data_format=input_data_format
+                )
 
+            all_images.append(image)
         images = [
-            to_channel_dimension_format(image, data_format, input_channel_dim=input_data_format) for image in images
+            to_channel_dimension_format(image, data_format, input_channel_dim=input_data_format)
+            for image in all_images
         ]
 
         data = {"pixel_values": images}

diff --git a/src/transformers/models/clip/image_processing_clip.py b/src/transformers/models/clip/image_processing_clip.py
@@ -319,31 +319,26 @@ def preprocess(
             # We assume that all images have the same channel dimension format.
             input_data_format = infer_channel_dimension_format(images[0])
 
-        if do_resize:
-            images = [
-                self.resize(image=image, size=size, resample=resample, input_data_format=input_data_format)
-                for image in images
-            ]
-
-        if do_center_crop:
-            images = [
-                self.center_crop(image=image, size=crop_size, input_data_format=input_data_format) for image in images
-            ]
-
-        if do_rescale:
-            images = [
-                self.rescale(image=image, scale=rescale_factor, input_data_format=input_data_format)
-                for image in images
-            ]
-
-        if do_normalize:
-            images = [
-                self.normalize(image=image, mean=image_mean, std=image_std, input_data_format=input_data_format)
-                for image in images
-            ]
+        all_images = []
+        for image in images:
+            if do_resize:
+                image = self.resize(image=image, size=size, resample=resample, input_data_format=input_data_format)
 
+            if do_center_crop:
+                image = self.center_crop(image=image, size=crop_size, input_data_format=input_data_format)
+
+            if do_rescale:
+                image = self.rescale(image=image, scale=rescale_factor, input_data_format=input_data_format)
+
+            if do_normalize:
+                image = self.normalize(
+                    image=image, mean=image_mean, std=image_std, input_data_format=input_data_format
+                )
+
+            all_images.append(image)
         images = [
-            to_channel_dimension_format(image, data_format, input_channel_dim=input_data_format) for image in images
+            to_channel_dimension_format(image, data_format, input_channel_dim=input_data_format)
+            for image in all_images
         ]
 
         data = {"pixel_values": images}

diff --git a/src/transformers/models/deit/image_processing_deit.py b/src/transformers/models/deit/image_processing_deit.py
@@ -270,31 +270,26 @@ def preprocess(
             # We assume that all images have the same channel dimension format.
             input_data_format = infer_channel_dimension_format(images[0])
 
-        if do_resize:
-            images = [
-                self.resize(image=image, size=size, resample=resample, input_data_format=input_data_format)
-                for image in images
-            ]
+        all_images = []
+        for image in images:
+            if do_resize:
+                image = self.resize(image=image, size=size, resample=resample, input_data_format=input_data_format)
 
-        if do_center_crop:
-            images = [
-                self.center_crop(image=image, size=crop_size, input_data_format=input_data_format) for image in images
-            ]
+            if do_center_crop:
+                image = self.center_crop(image=image, size=crop_size, input_data_format=input_data_format)
 
-        if do_rescale:
-            images = [
-                self.rescale(image=image, scale=rescale_factor, input_data_format=input_data_format)
-                for image in images
-            ]
+            if do_rescale:
+                image = self.rescale(image=image, scale=rescale_factor, input_data_format=input_data_format)
 
-        if do_normalize:
-            images = [
-                self.normalize(image=image, mean=image_mean, std=image_std, input_data_format=input_data_format)
-                for image in images
-            ]
+            if do_normalize:
+                image = self.normalize(
+                    image=image, mean=image_mean, std=image_std, input_data_format=input_data_format
+                )
 
+            all_images.append(image)
         images = [
-            to_channel_dimension_format(image, data_format, input_channel_dim=input_data_format) for image in images
+            to_channel_dimension_format(image, data_format, input_channel_dim=input_data_format)
+            for image in all_images
         ]
 
         data = {"pixel_values": images}

diff --git a/src/transformers/models/deprecated/vit_hybrid/image_processing_vit_hybrid.py b/src/transformers/models/deprecated/vit_hybrid/image_processing_vit_hybrid.py
@@ -312,31 +312,26 @@ def preprocess(
             # We assume that all images have the same channel dimension format.
             input_data_format = infer_channel_dimension_format(images[0])
 
-        if do_resize:
-            images = [
-                self.resize(image=image, size=size, resample=resample, input_data_format=input_data_format)
-                for image in images
-            ]
+        all_images = []
+        for image in images:
+            if do_resize:
+                image = self.resize(image=image, size=size, resample=resample, input_data_format=input_data_format)
 
-        if do_center_crop:
-            images = [
-                self.center_crop(image=image, size=crop_size, input_data_format=input_data_format) for image in images
-            ]
+            if do_center_crop:
+                image = self.center_crop(image=image, size=crop_size, input_data_format=input_data_format)
 
-        if do_rescale:
-            images = [
-                self.rescale(image=image, scale=rescale_factor, input_data_format=input_data_format)
-                for image in images
-            ]
+            if do_rescale:
+                image = self.rescale(image=image, scale=rescale_factor, input_data_format=input_data_format)
 
-        if do_normalize:
-            images = [
-                self.normalize(image=image, mean=image_mean, std=image_std, input_data_format=input_data_format)
-                for image in images
-            ]
+            if do_normalize:
+                image = self.normalize(
+                    image=image, mean=image_mean, std=image_std, input_data_format=input_data_format
+                )
 
+            all_images.append(image)
         images = [
-            to_channel_dimension_format(image, data_format, input_channel_dim=input_data_format) for image in images
+            to_channel_dimension_format(image, data_format, input_channel_dim=input_data_format)
+            for image in all_images
         ]
 
         data = {"pixel_values": images}

diff --git a/src/transformers/models/llava_next/image_processing_llava_next.py b/src/transformers/models/llava_next/image_processing_llava_next.py
@@ -409,31 +409,26 @@ def _preprocess(
         """
         images = make_list_of_images(images)
 
-        if do_resize:
-            images = [
-                self.resize(image=image, size=size, resample=resample, input_data_format=input_data_format)
-                for image in images
-            ]
-
-        if do_center_crop:
-            images = [
-                self.center_crop(image=image, size=crop_size, input_data_format=input_data_format) for image in images
-            ]
-
-        if do_rescale:
-            images = [
-                self.rescale(image=image, scale=rescale_factor, input_data_format=input_data_format)
-                for image in images
-            ]
-
-        if do_normalize:
-            images = [
-                self.normalize(image=image, mean=image_mean, std=image_std, input_data_format=input_data_format)
-                for image in images
-            ]
+        all_images = []
+        for image in images:
+            if do_resize:
+                image = self.resize(image=image, size=size, resample=resample, input_data_format=input_data_format)
+
+            if do_center_crop:
+                image = self.center_crop(image=image, size=crop_size, input_data_format=input_data_format)
+
+            if do_rescale:
+                image = self.rescale(image=image, scale=rescale_factor, input_data_format=input_data_format)
+
+            if do_normalize:
+                image = self.normalize(
+                    image=image, mean=image_mean, std=image_std, input_data_format=input_data_format
+                )
 
+            all_images.append(image)
         images = [
-            to_channel_dimension_format(image, data_format, input_channel_dim=input_data_format) for image in images
+            to_channel_dimension_format(image, data_format, input_channel_dim=input_data_format)
+            for image in all_images
         ]
 
         return images

diff --git a/src/transformers/models/llava_next_video/image_processing_llava_next_video.py b/src/transformers/models/llava_next_video/image_processing_llava_next_video.py
@@ -272,31 +272,26 @@ def _preprocess(
             # We assume that all images have the same channel dimension format.
             input_data_format = infer_channel_dimension_format(images[0])
 
-        if do_resize:
-            images = [
-                self.resize(image=image, size=size, resample=resample, input_data_format=input_data_format)
-                for image in images
-            ]
-
-        if do_center_crop:
-            images = [
-                self.center_crop(image=image, size=crop_size, input_data_format=input_data_format) for image in images
-            ]
-
-        if do_rescale:
-            images = [
-                self.rescale(image=image, scale=rescale_factor, input_data_format=input_data_format)
-                for image in images
-            ]
-
-        if do_normalize:
-            images = [
-                self.normalize(image=image, mean=image_mean, std=image_std, input_data_format=input_data_format)
-                for image in images
-            ]
+        all_images = []
+        for image in images:
+            if do_resize:
+                image = self.resize(image=image, size=size, resample=resample, input_data_format=input_data_format)
 
+            if do_center_crop:
+                image = self.center_crop(image=image, size=crop_size, input_data_format=input_data_format)
+
+            if do_rescale:
+                image = self.rescale(image=image, scale=rescale_factor, input_data_format=input_data_format)
+
+            if do_normalize:
+                image = self.normalize(
+                    image=image, mean=image_mean, std=image_std, input_data_format=input_data_format
+                )
+
+            all_images.append(image)
         images = [
-            to_channel_dimension_format(image, data_format, input_channel_dim=input_data_format) for image in images
+            to_channel_dimension_format(image, data_format, input_channel_dim=input_data_format)
+            for image in all_images
         ]
 
         return images