More updates for user-study

data_prep/proofread_consensus_annotations.py

@@ -0,0 +1,55 @@
+import os
+from glob import glob
+
+import imageio.v3 as imageio
+import napari
+from skimage.measure import label
+
+
+IMAGE_ROOT = "../data/for_annotation"
+LABEL_ROOT = "../data/consensus_labels/automatic"
+OUT_ROOT = "../data/consensus_labels/proofread"
+
+
+def proofread_split(split, skip_proofread):
+    image_folder = os.path.join(IMAGE_ROOT, f"split{split}")
+    split_images = sorted(glob(os.path.join(image_folder, "*.tif")))
+
+    label_folder = os.path.join(LABEL_ROOT, f"split{split}")
+    split_labels = sorted(glob(os.path.join(label_folder, "*.tif")))
+    assert len(split_images) == len(split_labels)
+
+    out_folder = os.path.join(OUT_ROOT, f"split{split}")
+    os.makedirs(out_folder, exist_ok=True)
+
+    for im_path, label_path in zip(split_images, split_labels):
+        im = imageio.imread(im_path)
+        labels = imageio.imread(label_path)
+
+        out_path = os.path.join(out_folder, os.path.basename(im_path))
+        if os.path.exists(out_path) and skip_proofread:
+            print("Skipping already proof-read file", out_path)
+            continue
+
+        v = napari.Viewer()
+        v.add_image(im)
+        v.add_labels(labels)
+
+        @v.bind_key("s")
+        def save_labels(v):
+            labels = v.layers["labels"].data
+            labels = label(labels)
+            print("Saving labels to", out_path)
+            imageio.imwrite(out_path, labels, compression="zlib")
+
+        napari.run()
+
+
+def main():
+    skip_proofread = True
+    for split in [1, 2, 3]:
+        proofread_split(split, skip_proofread=skip_proofread)
+
+
+if __name__ == "__main__":
+    main()

export_for_figure.py

@@ -0,0 +1,8 @@
+import os
+import imageio.v3 as imageio
+
+import h5py
+
+root = ""
+image_name = "VID532_H12_1_11d16h00m.tif"
+

summary_evaluation.py

@@ -12,15 +12,22 @@
 
 ANN_ROOT = "data/annotations"
 ANNOTATORS = ["anwai", "caro", "constantin", "luca", "marei"]
-# TODO use the proofread ones instead
-CONSENSUS_ROOT = "data/consensus_labels/automatic"
+CONSENSUS_ROOT = "data/consensus_labels/proofread"
 VERSION_TO_SPLIT = {
     "v1": 1, "v2": 1, "v3": 1, "v4": 1,
     "v5": 2, "v6": 2,
     "v7": 3, "v8": 3,
 }
 
 
+def _size_filter(labels, min_size):
+    if min_size is not None:
+        ids, sizes = np.unique(labels, return_counts=True)
+        filter_ids = ids[sizes < min_size]
+        labels[np.isin(labels, filter_ids)] = 0
+    return labels
+
+
 def _load_labels(version, name, im_name, min_size=None):
     # Cellpose
     if version in [4, 6, 8]:
@@ -30,10 +37,7 @@ def _load_labels(version, name, im_name, min_size=None):
         label_path = os.path.join(ANN_ROOT, f"v{version}", name, f"{im_name}.tif")
         labels = imageio.imread(label_path)
     labels = label(labels)
-    if min_size is not None:
-        ids, sizes = np.unique(labels, return_counts=True)
-        filter_ids = ids[sizes < min_size]
-        labels[np.isin(labels, filter_ids)] = 0
+    labels = _size_filter(labels, min_size)
     return labels
 
 
@@ -181,6 +185,8 @@ def evaluate_annotation_quality():
         "msa_ann_dev": [],
     }
 
+    min_size = 50
+
     for v in range(1, 9):
         version = f"v{v}"
         split = VERSION_TO_SPLIT[version]
@@ -190,9 +196,11 @@ def evaluate_annotation_quality():
 
         for label_path in consensus_labels:
             consensus = imageio.imread(label_path)
+            consensus = _size_filter(consensus, min_size)
+
             im_name = Path(label_path).stem
             for name in ANNOTATORS:
-                seg = _load_labels(v, name, im_name, min_size=50)
+                seg = _load_labels(v, name, im_name, min_size=min_size)
                 this_msa = mean_segmentation_accuracy(seg, consensus)
                 msas.append(this_msa)
 
@@ -247,8 +255,8 @@ def evaluate_generalization():
 
 
 def get_main_summary():
-    # summary_quality = evaluate_annotation_quality()
-    # return summary_quality
+    summary_quality = evaluate_annotation_quality()
+    return summary_quality
 
     summary_time = evaluate_annotation_times()
     summary_quality = evaluate_annotation_quality()
@@ -263,7 +271,7 @@ def get_main_summary():
 def main():
     summary = get_main_summary()
     print(summary)
-    summary.to_excel("./result_summary.xlsx")
+    # summary.to_excel("./result_summary.xlsx")
 
     # Times are in seconds / image EXCEPT Training time, which is in hours
     # additional time evaluation for runtimes.