[ENG-480] Simplify path by default, change API to num points

MANIFEST.in

@@ -13,4 +13,5 @@ recursive-include docs *.rst
 recursive-include labelbox *.py
 include labelbox/exporters/pascal_voc_writer/templates/annotation.xml
 recursive-include tests *.json
+recursive-include tests *.png
 recursive-include tests *.py

labelbox/__init__.py

@@ -1,3 +1,3 @@
 "The Labelbox python package."
 
-__version__ = '0.0.2'
+__version__ = '0.0.3'

labelbox/predictions/__init__.py

@@ -9,7 +9,7 @@
 def vectorize_to_v4_label(
         segmentation_map,
         legend: Dict[int, str],
-        epsilon: Optional[float] = None) -> DefaultDict[str, List[dict]]:
+        max_num_points: Optional[int] = 50) -> DefaultDict[str, List[dict]]:
     """Converts a segmentation map into polygons.
 
     Given a raster pixel wise array of predictions in `segmentation_map`,
@@ -25,10 +25,8 @@ def vectorize_to_v4_label(
         legend: A dictonary mapping pixel values
             used in `segmentation_map` to semantic
             class names.
-        epsilon: An optional argument, if present
-            controls the amount of path simplification.
-            Start with 1.0 and decrease to 0.01, 0.001.
-            No simplification occurs if absent.
+        max_num_points: The maximum number of points in the simplified path.
+            If `None`, then no path simplification is performed.
 
     Returns:
         A dictionary suitable for use as a `prediction`
@@ -43,8 +41,12 @@ class names.
         if pixel_value in legend and pixel_value is not 0:
             xy_list = polygon['coordinates'][0]
 
-            if epsilon:
+            if max_num_points:
+                epsilon = 0.001
                 xy_list = simplify_coords(xy_list, epsilon)
+                while len(xy_list) > max_num_points:
+                    epsilon *= 2
+                    xy_list = simplify_coords(xy_list, epsilon)
 
             geometry = []
             for point in xy_list:

tests/test_predictions.py

@@ -52,8 +52,24 @@ def test_vectorize_simplify():
     }
     segmentation_map = np.asarray(segmentation_map)
 
-    label = lbpreds.vectorize_to_v4_label(segmentation_map, legend, epsilon=None)
-    label_simple = lbpreds.vectorize_to_v4_label(segmentation_map, legend, epsilon=1.0)
+    label = lbpreds.vectorize_to_v4_label(segmentation_map, legend, max_num_points=None)
+    label_simple = lbpreds.vectorize_to_v4_label(segmentation_map, legend, max_num_points=10)
 
     # simplification reduces the number of points
     assert len(label['CLASS'][0]['geometry']) > len(label_simple['CLASS'][0]['geometry'])
+
+
+def test_vectorize_simplify_defaults(datadir):
+    with open(datadir.join('dog_prediction.png'), 'rb') as fp:
+        im = np.array(Image.open(fp))
+        legend = {
+                65535: 'Dog',
+        }
+        label = lbpreds.vectorize_to_v4_label(im, legend, max_num_points=None)
+
+        # a huge label with no simplification
+        assert len(label['Dog'][0]['geometry']) > 5000
+
+        # by default, simplifies to <= 50 points
+        label_simple = lbpreds.vectorize_to_v4_label(im, legend)
+        assert len(label_simple['Dog'][0]['geometry']) <= 50