Fix typos

README.md

@@ -47,19 +47,19 @@ processor = OneFormerProcessor.from_pretrained("shi-labs/oneformer_ade20k_swin_l
 model = OneFormerForUniversalSegmentation.from_pretrained("shi-labs/oneformer_ade20k_swin_large")
 
 # Semantic Segmentation
-semantic_inputs = processor(images=image, ["semantic"] return_tensors="pt")
+semantic_inputs = processor(images=image, task_inputs=["semantic"], return_tensors="pt")
 semantic_outputs = model(**semantic_inputs)
 # pass through image_processor for postprocessing
 predicted_semantic_map = processor.post_process_semantic_segmentation(outputs, target_sizes=[image.size[::-1]])[0]
 
 # Instance Segmentation
-instance_inputs = processor(images=image, ["instance"] return_tensors="pt")
+instance_inputs = processor(images=image, task_inputs=["instance"], return_tensors="pt")
 instance_outputs = model(**instance_inputs)
 # pass through image_processor for postprocessing
 predicted_instance_map = processor.post_process_instance_segmentation(outputs, target_sizes=[image.size[::-1]])[0]["segmentation"]
 
 # Panoptic Segmentation
-panoptic_inputs = processor(images=image, ["panoptic"] return_tensors="pt")
+panoptic_inputs = processor(images=image, task_inputs=["panoptic"], return_tensors="pt")
 panoptic_outputs = model(**panoptic_inputs)
 # pass through image_processor for postprocessing
 predicted_semantic_map = processor.post_process_panoptic_segmentation(outputs, target_sizes=[image.size[::-1]])[0]["segmentation"]