matched colormap for labels & segmentations

src/napari_sam_labeling_tools/_sam_rf_segmentation_widget.py

@@ -319,6 +319,7 @@ def check_label_layers(self, event: Event):
                 # to handle layer's name change by user
                 layer.events.name.disconnect()
                 layer.events.name.connect(self.check_label_layers)
+                layer.colormap = colormaps.create_colormap(10)[0]
                 if "Segmentation" in layer.name:
                     self.prediction_layer_combo.addItem(layer.name)
                 else:
@@ -536,6 +537,8 @@ def predict_slice(self, rf_model, slice_index, img_h, img_w):
             -1, SAM.PATCH_CHANNELS + SAM.EMBEDDING_SIZE
         )
         predictions = rf_model.predict(features).astype(np.uint8)
+        # to match the segmentation colormap with the labels
+        predictions[predictions > 0] += 1
         segmentation_image = predictions.reshape(img_h, img_w)
 
         # check for postprocessing

src/napari_sam_labeling_tools/utils/colormaps.py

@@ -55,7 +55,7 @@ def get_colormap1():
     return cm
 
 
-def create_colormap(num_colors, bright=True, black_first=True, seed=4567):
+def create_colormap(num_colors, bright=True, black_first=True, seed=777):
     if num_colors < 10:
         num_colors = 10
 
@@ -65,13 +65,14 @@ def create_colormap(num_colors, bright=True, black_first=True, seed=4567):
         low = [0.0, 0.55, 0.9]
         high = 1.0
 
+    rng = np.random.default_rng(seed)
+    hues = np.linspace(start=low[0], stop=high, num=num_colors)
+    rng.shuffle(hues)
     hsv_colors = np.stack([
-        np.linspace(start=low[0], stop=high, num=num_colors),
+        hues,
         np.linspace(start=low[1], stop=high, num=num_colors),
         np.linspace(start=low[2], stop=high, num=num_colors)
     ], axis=1)
-    rng = np.random.default_rng(seed)
-    rng.shuffle(hsv_colors)
 
     rgba_colors = np.zeros((num_colors, 4))
     for i in range(num_colors):

src/napari_sam_labeling_tools/widgets/utils.py

@@ -1,6 +1,8 @@
 import napari
 import numpy as np
 
+from napari_sam_labeling_tools.utils import colormaps
+
 
 def get_layer(napari_viewer, name, layer_types):
     for layer in napari_viewer.layers:
@@ -21,6 +23,7 @@ def add_labels_layer(napari_viewer: napari.Viewer):
         for s, sc in zip(scene_size, scale)
     ]
     empty_labels = np.zeros(shape, dtype=np.uint8)
-    napari_viewer.add_labels(
+    layer = napari_viewer.add_labels(
         empty_labels, name="Labels", translate=np.array(corner), scale=scale
     )
+    layer.colormap = colormaps.create_colormap(10)[0]