Merge pull request #111 from Janelia-Trackathon-2023/prune-tracking-g…

…raph

Prune tracking graph API

src/traccuracy/__init__.py

@@ -7,6 +7,6 @@
     __version__ = "uninstalled"
 
 from ._run_metrics import run_metrics
-from ._tracking_graph import EdgeAttr, NodeAttr, TrackingGraph
+from ._tracking_graph import EdgeFlag, NodeFlag, TrackingGraph
 
-__all__ = ["TrackingGraph", "run_metrics", "NodeAttr", "EdgeAttr"]
+__all__ = ["TrackingGraph", "run_metrics", "NodeFlag", "EdgeFlag"]

src/traccuracy/matchers/_base.py

@@ -50,9 +50,9 @@ def compute_mapping(
         matched.matcher_info = self.info
 
         # Report matching performance
-        total_gt = len(matched.gt_graph.nodes())
+        total_gt = len(matched.gt_graph.nodes)
         matched_gt = len({m[0] for m in matched.mapping})
-        total_pred = len(matched.pred_graph.nodes())
+        total_pred = len(matched.pred_graph.nodes)
         matched_pred = len({m[1] for m in matched.mapping})
         logger.info(f"Matched {matched_gt} out of {total_gt} ground truth nodes.")
         logger.info(f"Matched {matched_pred} out of {total_pred} predicted nodes.")

src/traccuracy/matchers/_ctc.py

@@ -37,7 +37,7 @@ def _compute_mapping(self, gt_graph: TrackingGraph, pred_graph: TrackingGraph):
             traccuracy.matchers.Matched: Matched data object containing the CTC mapping
 
         Raises:
-            ValueError: GT and pred segmentations must be the same shape
+            ValueError: if GT and pred segmentations are None or are not the same shape
         """
         gt = gt_graph
         pred = pred_graph
@@ -46,6 +46,9 @@ def _compute_mapping(self, gt_graph: TrackingGraph, pred_graph: TrackingGraph):
         G_gt, mask_gt = gt, gt.segmentation
         G_pred, mask_pred = pred, pred.segmentation
 
+        if mask_gt is None or mask_pred is None:
+            raise ValueError("Segmentation is None, cannot perform matching")
+
         if mask_gt.shape != mask_pred.shape:
             raise ValueError("Segmentation shapes must match between gt and pred")
 

src/traccuracy/matchers/_iou.py

@@ -1,5 +1,7 @@
 from __future__ import annotations
 
+from typing import Hashable
+
 import numpy as np
 from tqdm import tqdm
 
@@ -46,6 +48,35 @@ def _match_nodes(gt, res, threshold=1):
     return gtcells, rescells
 
 
+def _construct_time_to_seg_id_map(
+    graph: TrackingGraph,
+) -> dict[int, dict[Hashable, Hashable]]:
+    """For each time frame in the graph, create a mapping from segmentation ids
+    (the ids in the segmentation array, stored in graph.label_key) to the
+    node ids (the ids of the TrackingGraph nodes).
+
+    Args:
+        graph(TrackingGraph): a tracking graph with a label_key on each node
+
+    Returns:
+      dict[int, dict[Hashable, Hashable]]: a dictionary from {time: {segmentation_id: node_id}}
+
+    Raises:
+        AssertionError: If two nodes in a time frame have the same segmentation_id
+    """
+    time_to_seg_id_map: dict[int, dict[Hashable, Hashable]] = {}
+    for node_id, data in graph.nodes(data=True):
+        time = data[graph.frame_key]
+        seg_id = data[graph.label_key]
+        seg_id_to_node_id_map = time_to_seg_id_map.get(time, {})
+        assert (
+            seg_id not in seg_id_to_node_id_map
+        ), f"Segmentation ID {seg_id} occurred twice in frame {time}."
+        seg_id_to_node_id_map[seg_id] = node_id
+        time_to_seg_id_map[time] = seg_id_to_node_id_map
+    return time_to_seg_id_map
+
+
 def match_iou(gt, pred, threshold=0.6):
     """Identifies pairs of cells between gt and pred that have iou > threshold
 
@@ -78,24 +109,19 @@ def match_iou(gt, pred, threshold=0.6):
     # Get overlaps for each frame
     frame_range = range(gt.start_frame, gt.end_frame)
     total = len(list(frame_range))
+
+    gt_time_to_seg_id_map = _construct_time_to_seg_id_map(gt)
+    pred_time_to_seg_id_map = _construct_time_to_seg_id_map(pred)
+
     for i, t in tqdm(enumerate(frame_range), desc="Matching frames", total=total):
         matches = _match_nodes(
             gt.segmentation[i], pred.segmentation[i], threshold=threshold
         )
-
         # Construct node id tuple for each match
-        for gt_id, pred_id in zip(*matches):
+        for gt_seg_id, pred_seg_id in zip(*matches):
             # Find node id based on time and segmentation label
-            gt_node = gt.get_nodes_with_attribute(
-                gt.label_key,
-                criterion=lambda x: x == gt_id,  # noqa
-                limit_to=gt.get_nodes_in_frame(t),
-            )[0]
-            pred_node = pred.get_nodes_with_attribute(
-                pred.label_key,
-                criterion=lambda x: x == pred_id,  # noqa
-                limit_to=pred.get_nodes_in_frame(t),
-            )[0]
+            gt_node = gt_time_to_seg_id_map[t][gt_seg_id]
+            pred_node = pred_time_to_seg_id_map[t][pred_seg_id]
             mapper.append((gt_node, pred_node))
     return mapper
 

src/traccuracy/metrics/_ctc.py

@@ -2,7 +2,7 @@
 
 from typing import TYPE_CHECKING
 
-from traccuracy._tracking_graph import EdgeAttr, NodeAttr
+from traccuracy._tracking_graph import EdgeFlag, NodeFlag
 from traccuracy.track_errors._ctc import evaluate_ctc_events
 
 from ._base import Metric
@@ -38,14 +38,14 @@ def _compute(self, data: Matched):
         evaluate_ctc_events(data)
 
         vertex_error_counts = {
-            "ns": len(data.pred_graph.get_nodes_with_flag(NodeAttr.NON_SPLIT)),
-            "fp": len(data.pred_graph.get_nodes_with_flag(NodeAttr.FALSE_POS)),
-            "fn": len(data.gt_graph.get_nodes_with_flag(NodeAttr.FALSE_NEG)),
+            "ns": len(data.pred_graph.get_nodes_with_flag(NodeFlag.NON_SPLIT)),
+            "fp": len(data.pred_graph.get_nodes_with_flag(NodeFlag.FALSE_POS)),
+            "fn": len(data.gt_graph.get_nodes_with_flag(NodeFlag.FALSE_NEG)),
         }
         edge_error_counts = {
-            "ws": len(data.pred_graph.get_edges_with_flag(EdgeAttr.WRONG_SEMANTIC)),
-            "fp": len(data.pred_graph.get_edges_with_flag(EdgeAttr.FALSE_POS)),
-            "fn": len(data.gt_graph.get_edges_with_flag(EdgeAttr.FALSE_NEG)),
+            "ws": len(data.pred_graph.get_edges_with_flag(EdgeFlag.WRONG_SEMANTIC)),
+            "fp": len(data.pred_graph.get_edges_with_flag(EdgeFlag.FALSE_POS)),
+            "fn": len(data.gt_graph.get_edges_with_flag(EdgeFlag.FALSE_NEG)),
         }
         error_sum = get_weighted_error_sum(
             vertex_error_counts,

src/traccuracy/metrics/_divisions.py

@@ -36,7 +36,7 @@
 
 from typing import TYPE_CHECKING
 
-from traccuracy._tracking_graph import NodeAttr
+from traccuracy._tracking_graph import NodeFlag
 from traccuracy.track_errors.divisions import _evaluate_division_events
 
 from ._base import Metric
@@ -90,15 +90,9 @@ def _compute(self, data: Matched):
         }
 
     def _calculate_metrics(self, g_gt, g_pred):
-        tp_division_count = len(
-            g_gt.get_nodes_with_attribute(NodeAttr.TP_DIV, lambda x: x)
-        )
-        fn_division_count = len(
-            g_gt.get_nodes_with_attribute(NodeAttr.FN_DIV, lambda x: x)
-        )
-        fp_division_count = len(
-            g_pred.get_nodes_with_attribute(NodeAttr.FP_DIV, lambda x: x)
-        )
+        tp_division_count = len(g_gt.get_nodes_with_flag(NodeFlag.TP_DIV))
+        fn_division_count = len(g_gt.get_nodes_with_flag(NodeFlag.FN_DIV))
+        fp_division_count = len(g_pred.get_nodes_with_flag(NodeFlag.FP_DIV))
 
         try:
             recall = tp_division_count / (tp_division_count + fn_division_count)

src/traccuracy/track_errors/_ctc.py

@@ -6,7 +6,7 @@
 
 from tqdm import tqdm
 
-from traccuracy._tracking_graph import EdgeAttr, NodeAttr
+from traccuracy._tracking_graph import EdgeFlag, NodeFlag
 
 if TYPE_CHECKING:
     from traccuracy.matchers import Matched
@@ -39,12 +39,12 @@ def get_vertex_errors(matched_data: Matched):
         logger.info("Node errors already calculated. Skipping graph annotation")
         return
 
-    comp_graph.set_node_attribute(list(comp_graph.nodes()), NodeAttr.TRUE_POS, False)
-    comp_graph.set_node_attribute(list(comp_graph.nodes()), NodeAttr.NON_SPLIT, False)
+    comp_graph.set_flag_on_all_nodes(NodeFlag.TRUE_POS, False)
+    comp_graph.set_flag_on_all_nodes(NodeFlag.NON_SPLIT, False)
 
     # will flip this when we come across the vertex in the mapping
-    comp_graph.set_node_attribute(list(comp_graph.nodes()), NodeAttr.FALSE_POS, True)
-    gt_graph.set_node_attribute(list(gt_graph.nodes()), NodeAttr.FALSE_NEG, True)
+    comp_graph.set_flag_on_all_nodes(NodeFlag.FALSE_POS, True)
+    gt_graph.set_flag_on_all_nodes(NodeFlag.FALSE_NEG, True)
 
     # we need to know how many computed vertices are "non-split", so we make
     # a mapping of gt vertices to their matched comp vertices
@@ -57,15 +57,16 @@ def get_vertex_errors(matched_data: Matched):
         gt_ids = dict_mapping[pred_id]
         if len(gt_ids) == 1:
             gid = gt_ids[0]
-            comp_graph.set_node_attribute(pred_id, NodeAttr.TRUE_POS, True)
-            comp_graph.set_node_attribute(pred_id, NodeAttr.FALSE_POS, False)
-            gt_graph.set_node_attribute(gid, NodeAttr.FALSE_NEG, False)
+            comp_graph.set_flag_on_node(pred_id, NodeFlag.TRUE_POS, True)
+            comp_graph.set_flag_on_node(pred_id, NodeFlag.FALSE_POS, False)
+            gt_graph.set_flag_on_node(gid, NodeFlag.FALSE_NEG, False)
         elif len(gt_ids) > 1:
-            comp_graph.set_node_attribute(pred_id, NodeAttr.NON_SPLIT, True)
-            comp_graph.set_node_attribute(pred_id, NodeAttr.FALSE_POS, False)
+            comp_graph.set_flag_on_node(pred_id, NodeFlag.NON_SPLIT, True)
+            comp_graph.set_flag_on_node(pred_id, NodeFlag.FALSE_POS, False)
             # number of split operations that would be required to correct the vertices
             ns_count += len(gt_ids) - 1
-            gt_graph.set_node_attribute(gt_ids, NodeAttr.FALSE_NEG, False)
+            for gt_id in gt_ids:
+                gt_graph.set_flag_on_node(gt_id, NodeFlag.FALSE_NEG, False)
 
     # Record presence of annotations on the TrackingGraph
     comp_graph.node_errors = True
@@ -87,34 +88,30 @@ def get_edge_errors(matched_data: Matched):
         get_vertex_errors(matched_data)
 
     induced_graph = comp_graph.get_subgraph(
-        comp_graph.get_nodes_with_flag(NodeAttr.TRUE_POS)
+        comp_graph.get_nodes_with_flag(NodeFlag.TRUE_POS)
     ).graph
 
-    comp_graph.set_edge_attribute(list(comp_graph.edges()), EdgeAttr.FALSE_POS, False)
-    comp_graph.set_edge_attribute(
-        list(comp_graph.edges()), EdgeAttr.WRONG_SEMANTIC, False
-    )
-    gt_graph.set_edge_attribute(list(gt_graph.edges()), EdgeAttr.FALSE_NEG, False)
+    comp_graph.set_flag_on_all_edges(EdgeFlag.FALSE_POS, False)
+    comp_graph.set_flag_on_all_edges(EdgeFlag.WRONG_SEMANTIC, False)
+    gt_graph.set_flag_on_all_edges(EdgeFlag.FALSE_NEG, False)
 
     gt_comp_mapping = {gt: comp for gt, comp in node_mapping if comp in induced_graph}
     comp_gt_mapping = {comp: gt for gt, comp in node_mapping if comp in induced_graph}
 
     # intertrack edges = connection between parent and daughter
     for graph in [comp_graph, gt_graph]:
         # Set to False by default
-        graph.set_edge_attribute(list(graph.edges()), EdgeAttr.INTERTRACK_EDGE, False)
+        graph.set_flag_on_all_edges(EdgeFlag.INTERTRACK_EDGE, False)
 
         for parent in graph.get_divisions():
-            for daughter in graph.get_succs(parent):
-                graph.set_edge_attribute(
-                    (parent, daughter), EdgeAttr.INTERTRACK_EDGE, True
+            for daughter in graph.graph.successors(parent):
+                graph.set_flag_on_edge(
+                    (parent, daughter), EdgeFlag.INTERTRACK_EDGE, True
                 )
 
         for merge in graph.get_merges():
-            for parent in graph.get_preds(merge):
-                graph.set_edge_attribute(
-                    (parent, merge), EdgeAttr.INTERTRACK_EDGE, True
-                )
+            for parent in graph.graph.predecessors(merge):
+                graph.set_flag_on_edge((parent, merge), EdgeFlag.INTERTRACK_EDGE, True)
 
     # fp edges - edges in induced_graph that aren't in gt_graph
     for edge in tqdm(induced_graph.edges, "Evaluating FP edges"):
@@ -124,32 +121,32 @@ def get_edge_errors(matched_data: Matched):
         target_gt_id = comp_gt_mapping[target]
 
         expected_gt_edge = (source_gt_id, target_gt_id)
-        if expected_gt_edge not in gt_graph.edges():
-            comp_graph.set_edge_attribute(edge, EdgeAttr.FALSE_POS, True)
+        if expected_gt_edge not in gt_graph.edges:
+            comp_graph.set_flag_on_edge(edge, EdgeFlag.FALSE_POS, True)
         else:
             # check if semantics are correct
-            is_parent_gt = gt_graph.edges()[expected_gt_edge][EdgeAttr.INTERTRACK_EDGE]
-            is_parent_comp = comp_graph.edges()[edge][EdgeAttr.INTERTRACK_EDGE]
+            is_parent_gt = gt_graph.edges[expected_gt_edge][EdgeFlag.INTERTRACK_EDGE]
+            is_parent_comp = comp_graph.edges[edge][EdgeFlag.INTERTRACK_EDGE]
             if is_parent_gt != is_parent_comp:
-                comp_graph.set_edge_attribute(edge, EdgeAttr.WRONG_SEMANTIC, True)
+                comp_graph.set_flag_on_edge(edge, EdgeFlag.WRONG_SEMANTIC, True)
 
     # fn edges - edges in gt_graph that aren't in induced graph
-    for edge in tqdm(gt_graph.edges(), "Evaluating FN edges"):
+    for edge in tqdm(gt_graph.edges, "Evaluating FN edges"):
         source, target = edge[0], edge[1]
         # this edge is adjacent to an edge we didn't detect, so it definitely is an fn
         if (
-            gt_graph.nodes()[source][NodeAttr.FALSE_NEG]
-            or gt_graph.nodes()[target][NodeAttr.FALSE_NEG]
+            gt_graph.nodes[source][NodeFlag.FALSE_NEG]
+            or gt_graph.nodes[target][NodeFlag.FALSE_NEG]
         ):
-            gt_graph.set_edge_attribute(edge, EdgeAttr.FALSE_NEG, True)
+            gt_graph.set_flag_on_edge(edge, EdgeFlag.FALSE_NEG, True)
             continue
 
         source_comp_id = gt_comp_mapping[source]
         target_comp_id = gt_comp_mapping[target]
 
         expected_comp_edge = (source_comp_id, target_comp_id)
         if expected_comp_edge not in induced_graph.edges:
-            gt_graph.set_edge_attribute(edge, EdgeAttr.FALSE_NEG, True)
+            gt_graph.set_flag_on_edge(edge, EdgeFlag.FALSE_NEG, True)
 
     gt_graph.edge_errors = True
     comp_graph.edge_errors = True