Fixing object detection with layoutlm (#20776)

* Fixing object detection with layoutlm.

* Fixup.
This commit is contained in:
Nicolas Patry 2022-12-15 18:46:43 +01:00 committed by GitHub
parent 8891193e83
commit fca66abe2a
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 9 additions and 12 deletions

View File

@ -378,7 +378,12 @@ NO_TOKENIZER_TASKS = set()
# any tokenizer/feature_extractor might be use for a given model so we cannot
# use the statically defined TOKENIZER_MAPPING and FEATURE_EXTRACTOR_MAPPING to
# see if the model defines such objects or not.
MULTI_MODEL_CONFIGS = {"SpeechEncoderDecoderConfig", "VisionEncoderDecoderConfig", "VisionTextDualEncoderConfig"}
MULTI_MODEL_CONFIGS = {
"SpeechEncoderDecoderConfig",
"VisionEncoderDecoderConfig",
"VisionTextDualEncoderConfig",
"LayoutLMConfig",
}
for task, values in SUPPORTED_TASKS.items():
if values["type"] == "text":
NO_FEATURE_EXTRACTOR_TASKS.add(task)

View File

@ -256,17 +256,9 @@ class ObjectDetectionPipelineTests(unittest.TestCase, metaclass=PipelineTestCase
"https://huggingface.co/spaces/impira/docquery/resolve/2359223c1837a7587402bda0f2643382a6eefeab/invoice.png"
)
self.assertEqual(
nested_simplify(outputs, decimals=4),
nested_simplify(outputs, decimals=3),
[
{
"score": 0.9982,
"label": "B-QUESTION",
"box": {"xmin": 654, "ymin": 165, "xmax": 719, "ymax": 719},
},
{
"score": 0.9982,
"label": "I-QUESTION",
"box": {"xmin": 691, "ymin": 202, "xmax": 735, "ymax": 735},
},
{"score": 0.998, "label": "B-QUESTION", "box": {"xmin": 462, "ymin": 234, "xmax": 508, "ymax": 249}},
{"score": 0.999, "label": "I-QUESTION", "box": {"xmin": 489, "ymin": 286, "xmax": 519, "ymax": 301}},
],
)