rollback edt as incompatible with conda

conda-build/conda_build_config.yml

@@ -4,6 +4,10 @@ python:
   - 3.9
   - 3.10
   - 3.11
+channels:
+  - open3d-admin
+  - defaults
+  - conda-forge
 dependencies:
   - python>=3.8
   - pytorch>=1.12.0
@@ -14,7 +18,6 @@ dependencies:
   - scikit-learn>1.0.1
   - scikit-image>0.19.2
   - scipy>=1.8.1
-  - edt>=2.3.0
   - pillow>10.0.0
   - open3d>=0.9.0
   - requests>2.28.0

conda-build/meta-v0.2.1.yaml

@@ -20,7 +20,6 @@ requirements:
         - scikit-learn>1.0.1
         - scikit-image>0.19.2
         - scipy>=1.8.1
-        - edt>=2.3.0
         - pillow>10.0.0
         - open3d>=0.9.0
         - requests>2.28.0

conda-build/meta.yaml

@@ -20,7 +20,6 @@ requirements:
         - scikit-learn>1.0.1
         - scikit-image>0.19.2
         - scipy>=1.8.1
-        - edt>=2.3.0
         - pillow>10.0.0
         - open3d>=0.9.0
         - requests>2.28.0

requirements.txt

@@ -11,7 +11,6 @@ imagecodecs>2021.11.00
 scikit-learn>1.0.1
 scikit-image>0.19.2
 scipy>=1.8.1
-edt>=2.3.0
 pillow>10.0.0
 
 # External file format reader

tardis_em/dist_pytorch/utils/build_point_cloud.py

@@ -18,8 +18,6 @@
 
 from tardis_em.dist_pytorch.utils.utils import VoxelDownSampling
 from tardis_em.utils.errors import TardisError
-from tardis_em.dist_pytorch.utils.utils import pc_median_dist
-from copy import deepcopy
 
 
 class BuildPointCloud:
@@ -129,67 +127,31 @@ def build_point_cloud(
 
             return e, e
 
-        if EDT:
-            import edt
+        """Skeletonization"""
+        if image.ndim == 2:
+            image_point = skeletonize(image)
+        elif as_2d:
+            image_point = np.zeros(image.shape, dtype=np.uint8)
 
-            """Calculate EDT and apply threshold based on predefine mask size"""
-            if image.ndim == 2:
-                image_edt = edt.edt(image)
-                image_edt = np.where(image_edt > mask_size, 1, 0)
-            else:
-                image_edt = np.zeros(image.shape, dtype=np.uint8)
-
-                if as_2d:
-                    for i in range(image_edt.shape[0]):
-                        df_edt = edt.edt(image[i, :])
-
-                        image_edt[i, :] = np.where(df_edt > mask_size, 1, 0)
-                else:
-                    image_edt = edt.edt(image)
-                    image_edt = np.where(image_edt > mask_size, 1, 0)
-
-            image_edt = image_edt.astype(np.uint8)
-
-            """Skeletonization"""
-            if image.ndim == 2:
-                image_point = skeletonize(image_edt)
-            elif as_2d:
-                image_point = np.zeros(image_edt.shape, dtype=np.uint8)
-
-                for i in range(image_point.shape[0]):
-                    image_point[i, :] = np.where(skeletonize(image_edt[i, :]), 1, 0)
-            else:
-                image_point = skeletonize_3d(image_edt)
-            image_point = np.where(image_point > 0)
-
-            """CleanUp to avoid memory loss"""
-            del image, image_edt
-        else:
-            """Skeletonization"""
-            if image.ndim == 2:
-                image_point = skeletonize(image)
-            elif as_2d:
-                image_point = np.zeros(image.shape, dtype=np.uint8)
-
-                for i in range(image_point.shape[0]):
-                    image_point[i, :] = np.where(skeletonize(image[i, :]), 1, 0)
-            else:
-                image_point = skeletonize_3d(image)
-            image_point = np.where(image_point > 0)
-
-            """CleanUp to avoid memory loss"""
-            del image
-
-        """Output point cloud [X x Y x Z]"""
-        if len(image_point) == 2:
-            """If 2D bring artificially Z dim == 0"""
-            coordinates_HD = np.stack(
-                (image_point[1], image_point[0], np.zeros(image_point[0].shape))
-            ).T
+            for i in range(image_point.shape[0]):
+                image_point[i, :] = np.where(skeletonize(image[i, :]), 1, 0)
         else:
-            coordinates_HD = np.stack(
-                (image_point[2], image_point[1], image_point[0])
-            ).T
+            image_point = skeletonize_3d(image)
+        image_point = np.where(image_point > 0)
+
+        """CleanUp to avoid memory loss"""
+        del image
+
+    """Output point cloud [X x Y x Z]"""
+    if len(image_point) == 2:
+        """If 2D bring artificially Z dim == 0"""
+        coordinates_HD = np.stack(
+            (image_point[1], image_point[0], np.zeros(image_point[0].shape))
+        ).T
+    else:
+        coordinates_HD = np.stack(
+            (image_point[2], image_point[1], image_point[0])
+        ).T
 
         """CleanUp to avoid memory loss"""
         del image_point