def test_crop_in_batch():
uri = "http://farm2.staticflickr.com/1129/4726871278_4dd241a03a_z.jpg"
img = Image(uri)
data = img.to_numpy()
patches = img.crop(
[Box2d(10, 10, 30, 30),
Box2d(15, 15, 35, 35),
Box2d(20, 20, 40, 40)])
assert len(patches) == 3
assert np.array_equal(patches[0].to_numpy(), data[10:30, 10:30, :])
assert np.array_equal(patches[1].to_numpy(), data[15:35, 15:35, :])
assert np.array_equal(patches[2].to_numpy(), data[20:40, 20:40, :])
def deserialize(self, datum: Row) -> "rikai.types.geometry.Box2d":
from rikai.types.geometry import Box2d
if len(datum) < 4:
logger.error(f"Deserialize box2d: not sufficient data: {datum}")
return Box2d(*datum[:4])
def box2d_from_center(coords) -> Box2d:
"""Build a Box2d from a center-point based coordinate array:
``[center_x, center_y, width, height]``.
See Also
--------
:py:meth:`rikai.types.geometry.Box2d.from_center`
"""
return Box2d.from_center(coords[0], coords[1], coords[2], coords[3])
def test_dataframe_with_udt():
data = [{"id": i, "box": Box2d(i, i + 1, i + 2, i + 3)} for i in range(10)]
df = pd.DataFrame(data)
dataset = PandasDataset(df)
assert len(dataset) == 10
expected = [
{"id": i, "box": np.array([i, i + 1, i + 2, i + 3])} for i in range(10)
]
for i in range(10):
row = dataset[i]
assert row["id"] == expected[i]["id"]
assert np.array_equal(row["box"], expected[i]["box"])
def box2d_from_top_left(coords) -> Box2d:
"""Build :py:class:`Box2d` from a top-left based coordinate array:
``[x0, y0, width, height]``
Example
-------
`Coco dataset <https://cocodataset.org/>`_ is one public dataset that use
top-left coordinates
>>> #! pyspark --packages ai.eto:rikai_2.12:0.0.1
>>> import json
>>> from rikai.spark.functions.geometry import box2d_from_top_left
>>>
>>> with open("coco_sample/annotations/train_sample.json") as fobj:
... coco = json.load(fobj)
>>> anno_df = (
... spark
... .createDataFrame(coco["annotations"][:5])
... .withColumn("box2d", box2d_from_top_left("bbox"))
... )
>>> anno_df.show()
+--------------------+-----------+--------+--------------------+
| bbox|category_id|image_id| box2d|
+--------------------+-----------+--------+--------------------+
|[505.24, 0.0, 47....| 72| 318219|Box2d(x=505.24, y...|
|[470.68, 0.0, 45....| 72| 318219|Box2d(x=470.68, y...|
|[442.51, 0.0, 43....| 72| 318219|Box2d(x=442.51, y...|
|[380.74, 112.85, ...| 72| 554625|Box2d(x=380.74, y...|
|[339.13, 32.99, 3...| 72| 554625|Box2d(x=339.13, y...|
+--------------------+-----------+--------+--------------------+
>>> anno_df.printSchema()
root
|-- bbox: array (nullable = true)
| |-- element: double (containsNull = true)
|-- category_id: long (nullable = true)
|-- image_id: long (nullable = true)
|-- box2d: box2d (nullable = true)
See Also
--------
:py:meth:`rikai.types.geometry.Box2d.from_top_left`.
"""
return Box2d.from_top_left(coords[0], coords[1], coords[2], coords[3])
def box2d(coords) -> Box2d:
"""Build a Box2d from ``[xmin,ymin,xmax,ymax]`` array."""
return Box2d(coords[0], coords[1], coords[2], coords[3])
def test_pandas_series_with_udt():
data = pd.Series([Box2d(i, i + 1, i + 2, i + 3) for i in range(10)])
dataset = PandasDataset(data)
assert len(dataset) == 10
for i in range(10):
assert np.array_equal(dataset[i], np.array([i, i + 1, i + 2, i + 3]))
def test_crop_image():
data = np.random.randint(0, 255, size=(100, 100), dtype=np.uint8)
im = Image.from_array(data)
patch = im.crop(Box2d(10, 10, 30, 30))
cropped_data = patch.to_numpy()
assert np.array_equal(cropped_data, data[10:30, 10:30])
def test_crop_real_image():
uri = "http://farm2.staticflickr.com/1129/4726871278_4dd241a03a_z.jpg"
img = Image(uri)
data = img.to_numpy()
patch = img.crop(Box2d(10, 10, 30, 30))
assert np.array_equal(patch.to_numpy(), data[10:30, 10:30, :])