WIP bufixing : fixed errors in inference + instance seg +

Author: C-Achard

src/napari_cellseg3d/model_instance_seg.py

@@ -21,7 +21,7 @@ def binary_connected(
         thres_small (int): size threshold of small objects to remove. Default: 128
         scale_factors (tuple): scale factors for resizing in :math:`(Z, Y, X)` order. Default: (1.0, 1.0, 1.0)
     """
-    semantic = volume[0]
+    semantic = np.squeeze(volume)
     foreground = semantic > thres  # int(255 * thres)
     segm = label(foreground)
     segm = remove_small_objects(segm, thres_small)
@@ -64,9 +64,9 @@ def binary_watershed(
         thres_objects (float): threshold for foreground objects. Default: 0.3
         thres_small (int): size threshold of small objects removal. Default: 10
         scale_factors (tuple): scale factors for resizing in :math:`(Z, Y, X)` order. Default: (1.0, 1.0, 1.0)
-        rem_seed_thres: threshold for small seeds removal. Default : 3
+        rem_seed_thres (int): threshold for small seeds removal. Default : 3
     """
-    semantic = volume[0]
+    semantic = np.squeeze(volume)
     seed_map = semantic > thres_seeding
     foreground = semantic > thres_objects
     seed = label(seed_map)

src/napari_cellseg3d/model_workers.py

@@ -197,11 +197,13 @@ def inference(self):
         self.log("\nChecking dimensions...")
         pad = utils.get_padding_dim(check)
         # print(pad)
+        dims =128
+        dims=64
 
         model = self.model_dict["class"].get_net()
         if self.model_dict["name"] == "SegResNet":
             model = self.model_dict["class"].get_net()(
-                input_image_size=[128, 128, 128],  # TODO FIX !
+                input_image_size=[dims,dims,dims],  # TODO FIX !
                 out_channels=1,
                 # dropout_prob=0.3,
             )
@@ -294,6 +296,7 @@ def inference(self):
 
                 out = post_process_transforms(out)
                 out = np.array(out).astype(np.float32)
+                out = np.squeeze(out)
 
                 # batch_len = out.shape[1]
                 # print("trying to check len")
@@ -486,7 +489,7 @@ def log_parameters(self):
         if self.weights_path is not None:
             self.log(f"Using weights from : {self.weights_path}")
 
-        self.log("\n")
+        # self.log("\n")
 
     def train(self):
         """Trains the Pytorch model for the given number of epochs, with the selected model and data,
@@ -716,6 +719,7 @@ def train(self):
         self.log_parameters()
 
         for epoch in range(self.max_epochs):
+            # self.log("\n")
             self.log("-" * 10)
             self.log(f"Epoch {epoch + 1}/{self.max_epochs}")
             if self.device.type == "cuda":

src/napari_cellseg3d/plugin_crop.py

@@ -227,11 +227,17 @@ def start(self):
             self.image_path, self.filetype, self.as_folder
         )
 
+        if len(self.image.shape) > 3:
+            self.image = np.squeeze(self.image)
+
         if self.crop_labels:
             self.label = utils.load_images(
                 self.label_path, self.filetype, self.as_folder
             )
 
+            if len(self.label.shape) > 3:
+                self.label = np.squeeze(self.label)
+
         vw = self._viewer
 
         vw.dims.ndisplay = 3
@@ -274,6 +280,11 @@ def add_crop_sliders(
         self._y = 0
         self._z = 0
 
+        print(f"Crop variables")
+        print(image_stack.shape)
+
+
+
         # define crop sizes and boundaries for the image
         crop_sizes = [self._crop_size_x, self._crop_size_y, self._crop_size_z]
         for i in range(len(crop_sizes)):
@@ -284,13 +295,10 @@ def add_crop_sliders(
         # shapez, shapey, shapex = image_stack.shape
         ends = np.asarray(image_stack.shape) - np.asarray(crop_sizes) + 1
 
-
-
         stepsizes = ends // 100
 
-        print(f"Crop variables")
         print(crop_sizes)
-        print(image_stack.shape)
+
         print(ends)
         print(stepsizes)
 

src/napari_cellseg3d/plugin_model_training.py

@@ -883,7 +883,7 @@ def on_yield(data, widget):
                 data["weights"],
                 os.path.join(
                     widget.results_path,
-                    f"latest_weights_aborted_training_{utils.get_time()}.pth",
+                    f"latest_weights_aborted_training_{utils.get_time_filepath()}.pth",
                 ),
             )
             widget.stop_requested = False