Refactor optimize merge (#125)

* refactor: optimized merge detection

* bug: removed subroutine

* removed test tqdm

* bug: close zip writer

* refactor: metrics in pandas df

* refactor: detect all sites

* bug: empty merge list

* bug: rescaling swcs

* bug: removed break

* updates

* minor updates

* updated stats

---------

Co-authored-by: anna-grim <[email protected]>

Author: anna-grim

src/segmentation_skeleton_metrics/skeleton_metric.py

@@ -456,14 +456,18 @@ def detect_splits(self):
 
             # Store results
             for process in as_completed(processes):
-                key, graph, split_percent = process.result()
-                n_edges = graph.number_of_edges()
-                n_gt_edges = graph.graph["n_edges"]
+                key, graph, n_split_edges = process.result()
+                n_before = graph.graph["n_edges"]
+                n_after = graph.number_of_edges()
+
+                n_missing = n_before - n_after
+                p_omit = 100 * (n_missing + n_split_edges) / n_before
+                p_split = 100 * n_split_edges / n_before
 
                 self.graphs[key] = graph
-                self.metrics.at[key, "% Omit"] = 1 - n_edges / n_gt_edges
+                self.metrics.at[key, "% Omit"] = p_omit
                 self.metrics.at[key, "# Splits"] = gutil.count_splits(graph)
-                self.metrics.loc[key, "% Split"] = split_percent
+                self.metrics.loc[key, "% Split"] = p_split
                 self.metrics.loc[key, "GT Run Length"] = graph.run_length
                 pbar.update(1)
 
@@ -765,7 +769,7 @@ def compute_edge_accuracy(self):
         for key in self.graphs:
             p_omit = self.metrics.loc[key, "% Omit"]
             p_merged = self.metrics.loc[key, "% Merged"]
-            self.metrics.loc[key, "Edge Accuracy"] = 1 - p_omit - p_merged
+            self.metrics.loc[key, "Edge Accuracy"] = 100 - p_omit - p_merged
 
     def compute_erl(self):
         """

src/segmentation_skeleton_metrics/split_detection.py

@@ -30,7 +30,7 @@ def run(process_id, graph):
 
     """
     # Initializations
-    split_cnt = 0
+    n_split_edges = 0
     source = get_leaf(graph)
     dfs_edges = deque(list(nx.dfs_edges(graph, source=source)))
     visited_edges = set()
@@ -47,15 +47,14 @@ def run(process_id, graph):
         label_j = int(graph.labels[j])
         if is_split(label_i, label_j):
             graph.remove_edge(i, j)
-            split_cnt += 1
+            n_split_edges += 1
         elif label_j == 0:
             check_misalignment(graph, visited_edges, i, j)
         visited_edges.add(frozenset({i, j}))
 
     # Finish
-    split_percent = split_cnt / graph.graph["n_edges"]
     graph.remove_nodes_with_label(0)
-    return process_id, graph, split_percent
+    return process_id, graph, n_split_edges
 
 
 def check_misalignment(graph, visited_edges, nb, root):