SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

Fix image segmentation pipeline errors, resolve backward compatibility issues (#19768)

Browse Source 
* Fix panoptic segmentation and pipeline
* Update ImageSegmentationPipeline tests and reenable test_small_model_pt
* Resolve backward compatibility issues
This commit is contained in:
Alara Dirik
2022-10-21 18:09:58 +03:00
committed by GitHub
parent b58d4f70f6
commit cca51aa151
5 changed files with 113 additions and 137 deletions
Download Patch File Download Diff File Expand all files Collapse all files

125
tests/pipelines/test_pipelines_image_segmentation.py
View File

@@ -81,7 +81,12 @@ class ImageSegmentationPipelineTests(unittest.TestCase, metaclass=PipelineTestCa
]
def run_pipeline_test(self, image_segmenter, examples):
outputs = image_segmenter("./tests/fixtures/tests_samples/COCO/000000039769.png", threshold=0.0)
outputs = image_segmenter(
"./tests/fixtures/tests_samples/COCO/000000039769.png",
threshold=0.0,
mask_threshold=0,
overlap_mask_area_threshold=0,
)
self.assertIsInstance(outputs, list)
n = len(outputs)
if isinstance(image_segmenter.model, (MaskFormerForInstanceSegmentation)):
@@ -97,15 +102,15 @@ class ImageSegmentationPipelineTests(unittest.TestCase, metaclass=PipelineTestCa
dataset = datasets.load_dataset("hf-internal-testing/fixtures_image_utils", "image", split="test")
# RGBA
outputs = image_segmenter(dataset[0]["file"])
outputs = image_segmenter(dataset[0]["file"], threshold=0.0, mask_threshold=0, overlap_mask_area_threshold=0)
m = len(outputs)
self.assertEqual([{"score": ANY(float, type(None)), "label": ANY(str), "mask": ANY(Image.Image)}] * m, outputs)
# LA
outputs = image_segmenter(dataset[1]["file"])
outputs = image_segmenter(dataset[1]["file"], threshold=0.0, mask_threshold=0, overlap_mask_area_threshold=0)
m = len(outputs)
self.assertEqual([{"score": ANY(float, type(None)), "label": ANY(str), "mask": ANY(Image.Image)}] * m, outputs)
# L
outputs = image_segmenter(dataset[2]["file"])
outputs = image_segmenter(dataset[2]["file"], threshold=0.0, mask_threshold=0, overlap_mask_area_threshold=0)
m = len(outputs)
self.assertEqual([{"score": ANY(float, type(None)), "label": ANY(str), "mask": ANY(Image.Image)}] * m, outputs)
@@ -126,7 +131,9 @@ class ImageSegmentationPipelineTests(unittest.TestCase, metaclass=PipelineTestCa
"./tests/fixtures/tests_samples/COCO/000000039769.png",
"./tests/fixtures/tests_samples/COCO/000000039769.png",
]
outputs = image_segmenter(batch, threshold=0.0, batch_size=batch_size)
outputs = image_segmenter(
batch, threshold=0.0, mask_threshold=0, overlap_mask_area_threshold=0, batch_size=batch_size
)
self.assertEqual(len(batch), len(outputs))
self.assertEqual(len(outputs[0]), n)
self.assertEqual(
@@ -152,55 +159,29 @@ class ImageSegmentationPipelineTests(unittest.TestCase, metaclass=PipelineTestCa
model = AutoModelForImageSegmentation.from_pretrained(model_id)
feature_extractor = AutoFeatureExtractor.from_pretrained(model_id)
image_segmenter = ImageSegmentationPipeline(
model=model,
feature_extractor=feature_extractor,
task="semantic",
threshold=0.0,
overlap_mask_area_threshold=0.0,
)
image_segmenter = ImageSegmentationPipeline(model=model, feature_extractor=feature_extractor)
outputs = image_segmenter(
"http://images.cocodataset.org/val2017/000000039769.jpg",
subtask="panoptic",
threshold=0.0,
mask_threshold=0.0,
overlap_mask_area_threshold=0.0,
)
# Shortening by hashing
for o in outputs:
o["mask"] = mask_to_test_readable(o["mask"])
# This is extremely brittle, and those values are made specific for the CI.
self.assertEqual(
nested_simplify(outputs, decimals=4),
[
{
"label": "LABEL_88",
"mask": {"hash": "7f0bf661a4", "shape": (480, 640), "white_pixels": 3},
"score": None,
},
{
"label": "LABEL_101",
"mask": {"hash": "10ab738dc9", "shape": (480, 640), "white_pixels": 8948},
"score": None,
},
{
"score": 0.004,
"label": "LABEL_215",
"mask": {"hash": "b431e0946c", "shape": (480, 640), "white_pixels": 298249},
"score": None,
"mask": {"hash": "a01498ca7c", "shape": (480, 640), "white_pixels": 307200},
},
]
# Temporary: Keeping around the old values as they might provide useful later
# [
# {
# "score": 0.004,
# "label": "LABEL_215",
# "mask": {"hash": "34eecd16bb", "shape": (480, 640), "white_pixels": 0},
# },
# {
# "score": 0.004,
# "label": "LABEL_215",
# "mask": {"hash": "34eecd16bb", "shape": (480, 640), "white_pixels": 0},
# },
# ],
],
)
outputs = image_segmenter(
@@ -209,6 +190,8 @@ class ImageSegmentationPipelineTests(unittest.TestCase, metaclass=PipelineTestCa
"http://images.cocodataset.org/val2017/000000039769.jpg",
],
threshold=0.0,
mask_threshold=0.0,
overlap_mask_area_threshold=0.0,
)
for output in outputs:
for o in output:
@@ -219,62 +202,18 @@ class ImageSegmentationPipelineTests(unittest.TestCase, metaclass=PipelineTestCa
[
[
{
"label": "LABEL_88",
"mask": {"hash": "7f0bf661a4", "shape": (480, 640), "white_pixels": 3},
"score": None,
},
{
"label": "LABEL_101",
"mask": {"hash": "10ab738dc9", "shape": (480, 640), "white_pixels": 8948},
"score": None,
},
{
"score": 0.004,
"label": "LABEL_215",
"mask": {"hash": "b431e0946c", "shape": (480, 640), "white_pixels": 298249},
"score": None,
"mask": {"hash": "a01498ca7c", "shape": (480, 640), "white_pixels": 307200},
},
],
[
{
"label": "LABEL_88",
"mask": {"hash": "7f0bf661a4", "shape": (480, 640), "white_pixels": 3},
"score": None,
},
{
"label": "LABEL_101",
"mask": {"hash": "10ab738dc9", "shape": (480, 640), "white_pixels": 8948},
"score": None,
},
{
"score": 0.004,
"label": "LABEL_215",
"mask": {"hash": "b431e0946c", "shape": (480, 640), "white_pixels": 298249},
"score": None,
"mask": {"hash": "a01498ca7c", "shape": (480, 640), "white_pixels": 307200},
},
]
# [
# {
# "score": 0.004,
# "label": "LABEL_215",
# "mask": {"hash": "34eecd16bb", "shape": (480, 640), "white_pixels": 0},
# },
# {
# "score": 0.004,
# "label": "LABEL_215",
# "mask": {"hash": "34eecd16bb", "shape": (480, 640), "white_pixels": 0},
# },
# ],
# [
# {
# "score": 0.004,
# "label": "LABEL_215",
# "mask": {"hash": "34eecd16bb", "shape": (480, 640), "white_pixels": 0},
# },
# {
# "score": 0.004,
# "label": "LABEL_215",
# "mask": {"hash": "34eecd16bb", "shape": (480, 640), "white_pixels": 0},
# },
# ],
],
],
)
@@ -311,7 +250,7 @@ class ImageSegmentationPipelineTests(unittest.TestCase, metaclass=PipelineTestCa
outputs = image_segmenter(
"http://images.cocodataset.org/val2017/000000039769.jpg",
task="panoptic",
subtask="panoptic",
threshold=0,
overlap_mask_area_threshold=0.0,
)
@@ -361,7 +300,7 @@ class ImageSegmentationPipelineTests(unittest.TestCase, metaclass=PipelineTestCa
"http://images.cocodataset.org/val2017/000000039769.jpg",
"http://images.cocodataset.org/val2017/000000039769.jpg",
],
task="panoptic",
subtask="panoptic",
threshold=0.0,
overlap_mask_area_threshold=0.0,
)
@@ -448,7 +387,7 @@ class ImageSegmentationPipelineTests(unittest.TestCase, metaclass=PipelineTestCa
image_segmenter = pipeline("image-segmentation", model=model_id)
outputs = image_segmenter(
"http://images.cocodataset.org/val2017/000000039769.jpg", task="panoptic", threshold=0.999
"http://images.cocodataset.org/val2017/000000039769.jpg", subtask="panoptic", threshold=0.999
)
# Shortening by hashing
for o in outputs:
@@ -471,7 +410,7 @@ class ImageSegmentationPipelineTests(unittest.TestCase, metaclass=PipelineTestCa
)
outputs = image_segmenter(
"http://images.cocodataset.org/val2017/000000039769.jpg", task="panoptic", threshold=0.5
"http://images.cocodataset.org/val2017/000000039769.jpg", subtask="panoptic", threshold=0.5
)
for o in outputs:
@@ -521,7 +460,7 @@ class ImageSegmentationPipelineTests(unittest.TestCase, metaclass=PipelineTestCa
image = load_dataset("hf-internal-testing/fixtures_ade20k", split="test")
file = image[0]["file"]
outputs = image_segmenter(file, task="panoptic", threshold=threshold)
outputs = image_segmenter(file, subtask="panoptic", threshold=threshold)
# Shortening by hashing
for o in outputs: