Add image height and width to ONNX dynamic axes (#18915)

src/transformers/models/beit/configuration_beit.py

@@ -194,7 +194,7 @@ class BeitOnnxConfig(OnnxConfig):
     def inputs(self) -> Mapping[str, Mapping[int, str]]:
         return OrderedDict(
             [
-                ("pixel_values", {0: "batch", 1: "num_channels"}),
+                ("pixel_values", {0: "batch", 1: "num_channels", 2: "height", 3: "width"}),
             ]
         )
 

src/transformers/models/clip/configuration_clip.py

@@ -332,7 +332,7 @@ class CLIPOnnxConfig(OnnxConfig):
         return OrderedDict(
             [
                 ("input_ids", {0: "batch", 1: "sequence"}),
-                ("pixel_values", {0: "batch"}),
+                ("pixel_values", {0: "batch", 1: "num_channels", 2: "height", 3: "width"}),
                 ("attention_mask", {0: "batch", 1: "sequence"}),
             ]
         )

src/transformers/models/convnext/configuration_convnext.py

@@ -117,7 +117,7 @@ class ConvNextOnnxConfig(OnnxConfig):
     def inputs(self) -> Mapping[str, Mapping[int, str]]:
         return OrderedDict(
             [
-                ("pixel_values", {0: "batch", 1: "num_channels"}),
+                ("pixel_values", {0: "batch", 1: "num_channels", 2: "height", 3: "width"}),
             ]
         )
 

src/transformers/models/data2vec/configuration_data2vec_vision.py

@@ -193,7 +193,7 @@ class Data2VecVisionOnnxConfig(OnnxConfig):
     def inputs(self) -> Mapping[str, Mapping[int, str]]:
         return OrderedDict(
             [
-                ("pixel_values", {0: "batch", 1: "num_channels"}),
+                ("pixel_values", {0: "batch", 1: "num_channels", 2: "height", 3: "width"}),
             ]
         )
 

src/transformers/models/deit/configuration_deit.py

@@ -137,7 +137,7 @@ class DeiTOnnxConfig(OnnxConfig):
     def inputs(self) -> Mapping[str, Mapping[int, str]]:
         return OrderedDict(
             [
-                ("pixel_values", {0: "batch", 1: "num_channels"}),
+                ("pixel_values", {0: "batch", 1: "num_channels", 2: "height", 3: "width"}),
             ]
         )
 

src/transformers/models/detr/configuration_detr.py

@@ -223,7 +223,7 @@ class DetrOnnxConfig(OnnxConfig):
     def inputs(self) -> Mapping[str, Mapping[int, str]]:
         return OrderedDict(
             [
-                ("pixel_values", {0: "batch", 1: "num_channels"}),
+                ("pixel_values", {0: "batch", 1: "num_channels", 2: "height", 3: "width"}),
                 ("pixel_mask", {0: "batch"}),
             ]
         )

src/transformers/models/layoutlmv3/configuration_layoutlmv3.py

@@ -203,7 +203,7 @@ class LayoutLMv3OnnxConfig(OnnxConfig):
                     ("input_ids", {0: "batch", 1: "sequence"}),
                     ("attention_mask", {0: "batch", 1: "sequence"}),
                     ("bbox", {0: "batch", 1: "sequence"}),
-                    ("pixel_values", {0: "batch", 1: "sequence"}),
+                    ("pixel_values", {0: "batch", 1: "num_channels", 2: "height", 3: "width"}),
                 ]
             )
         else:

src/transformers/models/levit/configuration_levit.py

@@ -137,7 +137,7 @@ class LevitOnnxConfig(OnnxConfig):
     def inputs(self) -> Mapping[str, Mapping[int, str]]:
         return OrderedDict(
             [
-                ("pixel_values", {0: "batch", 1: "num_channels"}),
+                ("pixel_values", {0: "batch", 1: "num_channels", 2: "height", 3: "width"}),
             ]
         )
 

src/transformers/models/mobilevit/configuration_mobilevit.py

@@ -171,7 +171,7 @@ class MobileViTOnnxConfig(OnnxConfig):
 
     @property
     def inputs(self) -> Mapping[str, Mapping[int, str]]:
-        return OrderedDict([("pixel_values", {0: "batch", 1: "num_channels"})])
+        return OrderedDict([("pixel_values", {0: "batch", 1: "num_channels", 2: "height", 3: "width"})])
 
     @property
     def outputs(self) -> Mapping[str, Mapping[int, str]]:

src/transformers/models/resnet/configuration_resnet.py

@@ -105,7 +105,7 @@ class ResNetOnnxConfig(OnnxConfig):
     def inputs(self) -> Mapping[str, Mapping[int, str]]:
         return OrderedDict(
             [
-                ("pixel_values", {0: "batch", 1: "num_channels"}),
+                ("pixel_values", {0: "batch", 1: "num_channels", 2: "height", 3: "width"}),
             ]
         )
 

src/transformers/models/vit/configuration_vit.py

@@ -135,7 +135,7 @@ class ViTOnnxConfig(OnnxConfig):
     def inputs(self) -> Mapping[str, Mapping[int, str]]:
         return OrderedDict(
             [
-                ("pixel_values", {0: "batch", 1: "num_channels"}),
+                ("pixel_values", {0: "batch", 1: "num_channels", 2: "height", 3: "width"}),
             ]
         )