Adding preprocessor parameters (#427)

Prevent it from giving an exception with newer models.


https://huggingface.co/microsoft/table-transformer-structure-recognition-v1.1-all/discussions/1

---------

Co-authored-by: Yao You <[email protected]>

Author: ajjimeno

CHANGELOG.md

@@ -1,3 +1,7 @@
+## 1.0.2
+
+* adding parameter to table image preprocessor related to the image size
+
 ## 1.0.1
 
 * fix: moving the table transformer model to device when loading the model instead of once the model is loaded.

unstructured_inference/__version__.py

@@ -1 +1 @@
-__version__ = "1.0.1"  # pragma: no cover
+__version__ = "1.0.2"  # pragma: no cover

unstructured_inference/models/tables.py

@@ -62,6 +62,9 @@ def initialize(
         """Loads the donut model using the specified parameters"""
         self.device = device
         self.feature_extractor = DetrImageProcessor.from_pretrained(model)
+        # value not set in the configuration and needed for newer models
+        # https://huggingface.co/microsoft/table-transformer-structure-recognition-v1.1-all/discussions/1
+        self.feature_extractor.size["shortest_edge"] = 800
 
         try:
             logger.info("Loading the table structure model ...")