Upload processor

image_processor.py

@@ -23,7 +23,7 @@ from transformers.image_utils import (
 from transformers.utils import is_torch_tensor
 
 
-class FaceNetImageProcessor(BaseImageProcessor):
+class FaceSegformerImageProcessor(BaseImageProcessor):
     def __init__(self, **kwargs):
         super().__init__(**kwargs)
         self.image_size = kwargs.get("image_size", (224, 224))
@@ -172,11 +172,7 @@ class FaceNetImageProcessor(BaseImageProcessor):
             images = [
                 self.resize(
                     image=image,
-                    size={
-                        "shortest_edge": min(
-                            kwargs.get("image_size") or self.image_size
-                        )
-                    },
+                    size={"height": self.image_size[0], "width": self.image_size[1]},
                     resample=kwargs.get("resample") or self.resample,
                     input_data_format=input_data_format,
                 )
@@ -204,7 +200,7 @@ class FaceNetImageProcessor(BaseImageProcessor):
         data = {"pixel_values": images}
         return BatchFeature(data=data, tensor_type="pt")
 
-    # Copied from transformers.models.beit.image_processing_beit.BeitImageProcessor.post_process_semantic_segmentation with Beit->Segformer
+    # Copied from transformers.models.segformer.image_processing_segformer.SegformerImageProcessor.post_process_semantic_segmentation
     def post_process_semantic_segmentation(
         self, outputs, target_sizes: List[Tuple] = None
     ):
@@ -254,37 +250,3 @@ class FaceNetImageProcessor(BaseImageProcessor):
             ]
 
         return semantic_segmentation
-
-    # def post_process_instance_segmentation(
-    #     self,
-    #     outputs,
-    #     target_sizes: List[Tuple] = None,
-    #     **kwargs
-    # ) -> np.ndarray:
-    #     logits = outputs.logits
-    #     if target_sizes is not None:
-    #         if len(logits) != len(target_sizes):
-    #             raise ValueError(
-    #                 "Make sure that you pass in as many target sizes as the batch dimension of the logits"
-    #             )
-    #         # use target sizes to resize logits
-    #         resized_masks = []
-    #         for idx in range(len(logits)):
-    #             resized_mask = torch.nn.functional.interpolate(
-    #                 logits[idx].unsqueeze(dim=0),
-    #                 size=target_sizes[idx],
-    #                 mode="bilinear",
-    #                 align_corners=False,
-    #             )
-    #             resized_masks.append(
-    #                 resized_mask[0].softmax(dim=0).argmax(dim=0).to("cpu").numpy()
-    #             )
-
-    #         predicted_masks = np.array(resized_masks)
-
-    #     else:
-    #         predicted_masks = (
-    #             torch.argmax(torch.softmax(logits, dim=1), axis=1).to("cpu").numpy()
-    #         )
-
-    #     return predicted_masks

preprocessor_config.json

@@ -1,9 +1,9 @@
 {
   "auto_map": {
-    "AutoImageProcessor": "image_processor.FaceNetImageProcessor"
+    "AutoImageProcessor": "image_processor.FaceSegformerImageProcessor"
   },
   "data_format": "channels_first",
-  "image_processor_type": "FaceNetImageProcessor",
+  "image_processor_type": "FaceSegformerImageProcessor",
   "image_size": [
     224,
     224