Create volume annotations

webknossos/Changelog.md

@@ -15,6 +15,7 @@ For upgrade instructions, please check the respective _Breaking Changes_ section
 ### Breaking Changes
 
 ### Added
+- Added context manager `VolumeLayer.edit` for creating and modifying volume annotations. [#1340](https://github.com/scalableminds/webknossos-libs/pull/1340)
 
 ### Changed
 

webknossos/examples/WIP/merge_trees_at_closest_nodes.py

@@ -6,19 +6,19 @@
 
 import webknossos as wk
 
-nml = wk.Skeleton.load("trees-in-groups.nml")
+skeleton = wk.Skeleton.load("trees-in-groups.nml")
 
 # Probably we want to keep groups and normal trees in distinct collections (groups/trees).
 # For many use-cases a common view groups_and_trees would be great, but not here:
-for group in nml.groups.values():  # groups is a dict with the name as keys
+for group in skeleton.groups.values():  # groups is a dict with the name as keys
     min_distance_graph = G = nx.Graph()
     for (tree_idx_a, tree_a), (tree_idx_b, tree_b) in combinations(
         enumerate(group.flattened_trees()), 2
     ):
         pos_a = (
-            tree_a.get_node_positions() * nml.voxel_size
+            tree_a.get_node_positions() * skeleton.voxel_size
         )  # or tree_a.get_node_positions_nm?
-        pos_b = tree_b.get_node_positions() * nml.voxel_size
+        pos_b = tree_b.get_node_positions() * skeleton.voxel_size
         node_idx_a, node_idx_b, distance = wk.geometry.closest_pair(pos_a, pos_b)
         G.add_edge((tree_idx_a, node_idx_a), (tree_idx_b, node_idx_b), weight=distance)
     new_edges = nx.algorithms.tree.mst.minimum_spanning_edges()
@@ -35,7 +35,7 @@
     final_tree.name = group.name
     final_tree.group = None
 
-    del nml.groups[group.name]
+    del skeleton.groups[group.name]
     # or
     group.delete()
     # The latter only works if everything is double-linked.
@@ -44,4 +44,4 @@
     # to do the double-linking. Simply dict-like insertions can't work then:
     # nml["tree-name"] = Tree()
 
-nml.save("merged-trees.nml")
+skeleton.save("merged-trees.nml")

webknossos/examples/WIP/offline_merger_mode.py

@@ -4,21 +4,21 @@
 
 import webknossos as wk
 
-# A merger mode nml with every tree corresponding to a new merged segment is available.
+# A merger mode skeleton with every tree corresponding to a new merged segment is available.
 # All segments in which a node is placed should be merged and saved as a new dataset.
 
-# for local nml:
-nml = wk.open("merger-mode.nml")
+# for local skeleton:
+skeleton = wk.open("merger-mode.skeleton")
 # wk.Skeleton.load or wk.open_skeleton works, too (and is type-safe)
 
 # for online annotation:
-annotation = wk.Annotation.download(
-    "https://webknossos.org/annotations/Explorational/6114d9410100009f0096c640"
-)
-nml = annotation.skeleton
+skeleton = wk.Annotation.download(
+    "https://webknossos.org/annotations/Explorational/6114d9410100009f0096c640",
+    skip_volume_data=True,
+).skeleton
 # should this save anything to disk, or just happen in memory?
 
-dataset = wk.download(nml.dataset_name, organization=nml.dataset_organization)
+dataset = wk.download(skeleton.dataset_name, organization=skeleton.dataset_organization)
 # asks for auth token, persisted into .env or similar config file (maybe use xdg-path?)
 
 # sub-part access via dicts or dict-like classes
@@ -28,7 +28,7 @@
 
 segmentation_data = view.read()
 
-for tree in nml.trees():  # nml.trees() is a flattened iterator of all trees
+for tree in skeleton.trees:  # skeleton.trees() is a flattened iterator of all trees
     segment_ids_in_tree = set(
         segmentation_data[tuple(node.position - view.topleft)] for node in tree.nodes
     )

webknossos/examples/apply_merger_mode.py

@@ -12,8 +12,9 @@ def main() -> None:
     # Opening a merger mode annotation #
     ####################################
 
-    nml = wk.Annotation.download(
-        "https://webknossos.org/annotations/6748612b0100001101c81156"
+    skeleton = wk.Annotation.download(
+        "https://webknossos.org/annotations/6748612b0100001101c81156",
+        skip_volume_data=True,
     ).skeleton
 
     ###############################################
@@ -33,7 +34,7 @@ def main() -> None:
     ##############################
 
     segment_id_mapping = {}
-    for tree in nml.flattened_trees():
+    for tree in skeleton.flattened_trees():
         base = None
         for node in tree.nodes:
             segment_id = in_mag1.read(
@@ -44,7 +45,7 @@ def main() -> None:
             segment_id_mapping[segment_id] = base
 
     print(
-        f"Found {len(list(nml.flattened_trees()))} segment id groups with {len(segment_id_mapping)} nodes"
+        f"Found {len(list(skeleton.flattened_trees()))} segment id groups with {len(segment_id_mapping)} nodes"
     )
     print(segment_id_mapping)
 

webknossos/examples/learned_segmenter.py

@@ -1,6 +1,4 @@
-import os
 from functools import partial
-from tempfile import TemporaryDirectory
 
 import numpy as np
 from skimage import feature
@@ -19,7 +17,6 @@ def main() -> None:
     # Step 1: Read the training data from the annotation and the dataset's color
     # layer (the data will be streamed from WEBKNOSSOS to our local computer)
     training_data_bbox = annotation.user_bounding_boxes[0]  # type: ignore[index]
-    new_dataset_name = f"{annotation.dataset_name.replace(' ', '_')}_segmented"
     with wk.webknossos_context("https://webknossos.org"):
         dataset = annotation.get_remote_annotation_dataset()
 
@@ -58,28 +55,15 @@ def main() -> None:
     assert segmentation.max() < 256
     segmentation = segmentation.astype("uint8")
 
-    # Step 5: Bundle everything as a WEBKNOSSOS layer and upload to wK for viewing and further work
-    with TemporaryDirectory() as tempdir:
-        new_dataset = wk.Dataset(
-            tempdir, voxel_size=dataset.voxel_size, name=new_dataset_name
-        )
-        segmentation_layer = new_dataset.add_layer(
-            "segmentation",
-            wk.SEGMENTATION_CATEGORY,
-            dtype_per_channel=segmentation.dtype,
-            largest_segment_id=int(segmentation.max()),
-        )
+    # Step 5: Upload the segmentation to WEBKNOSSOS
+    print("Uploading segmentation…")
+    volume_layer = annotation.add_volume_layer("segmentation", dtype=segmentation.dtype)
+    with volume_layer.edit() as segmentation_layer:
         segmentation_layer.bounding_box = dataset.layers["color"].bounding_box
         segmentation_layer.add_mag(mag, compress=True).write(segmentation)
         segmentation_layer.downsample(sampling_mode="constant_z")
 
-        remote_ds = new_dataset.upload(
-            layers_to_link=[annotation.get_remote_base_dataset().layers["color"]]
-            if "PYTEST_CURRENT_TEST" not in os.environ
-            else None
-        )
-
-    url = remote_ds.url
+    url = annotation.upload()
     print(f"Successfully uploaded {url}")
 
 

webknossos/examples/skeleton_path_length.py

@@ -6,9 +6,7 @@
 def calculate_path_length(annotation_url: str, auth_token: str) -> None:
     with wk.webknossos_context(token=auth_token):
         # Download a annotation directly from the WEBKNOSSOS server
-        annotation = wk.Annotation.download(
-            annotation_url,
-        )
+        annotation = wk.Annotation.download(annotation_url, skip_volume_data=True)
 
         skeleton = annotation.skeleton
         voxel_size = annotation.voxel_size