oversample

src/napatrackmater/Trackvector.py

@@ -3389,7 +3389,9 @@ def plot_at_mitosis_time(matrix_directory, save_dir, dataset_name, channel):
             plt.show()
 
 
-def plot_histograms_for_groups(matrix_directory, save_dir, dataset_name, channel, name = 'all'):
+def plot_histograms_for_groups(
+    matrix_directory, save_dir, dataset_name, channel, name="all"
+):
 
     files = os.listdir(matrix_directory)
     sorted_files = natsorted(
@@ -3440,7 +3442,7 @@ def plot_histograms_for_groups(matrix_directory, save_dir, dataset_name, channel
 
 
 def plot_histograms_for_cell_type_groups(
-    matrix_directory, save_dir, dataset_name, channel, label_dict = None, name = 'all'
+    matrix_directory, save_dir, dataset_name, channel, label_dict=None, name="all"
 ):
 
     files = os.listdir(matrix_directory)
@@ -3470,11 +3472,13 @@ def plot_histograms_for_cell_type_groups(
                 file_path = os.path.join(matrix_directory, file_name)
                 cell_type = extract_celltype(file_name)
                 if label_dict is not None:
-                    cell_type_name = label_dict[cell_type]  
+                    cell_type_name = label_dict[cell_type]
                 else:
-                    cell_type_name = cell_type     
+                    cell_type_name = cell_type
                 data = np.load(file_path, allow_pickle=True)
-                sns.histplot(data, alpha=0.5, kde=True, label=f"Cell_Type: {cell_type_name}")
+                sns.histplot(
+                    data, alpha=0.5, kde=True, label=f"Cell_Type: {cell_type_name}"
+                )
 
             plt.xlabel("Value")
             plt.ylabel("Counts")
@@ -4042,7 +4046,6 @@ def inception_model_prediction(
     class_map,
     dynamic_model=None,
     shape_model=None,
-    num_samples=10,
     device="cpu",
 ):
     sub_dataframe = dataframe[dataframe["Track ID"] == track_id]
@@ -4051,13 +4054,13 @@ def inception_model_prediction(
 
     total_duration = sub_dataframe["Track Duration"].max()
 
-    def sample_subarrays(data, num_samples, tracklet_length, total_duration):
+    if sub_dataframe.shape[0] < tracklet_length:
+        return "UnClassified"
+
+    def sample_subarrays(data, tracklet_length, total_duration):
 
         max_start_index = total_duration - tracklet_length
-        if max_start_index > num_samples:
-            start_indices = random.sample(range(max_start_index), num_samples)
-        else:
-            start_indices = [0] * num_samples
+        start_indices = random.sample(range(max_start_index), max_start_index)
 
         subarrays = []
         for start_index in start_indices:
@@ -4070,10 +4073,10 @@ def sample_subarrays(data, num_samples, tracklet_length, total_duration):
         return subarrays
 
     sub_arrays_shape = sample_subarrays(
-        sub_dataframe_shape, num_samples, tracklet_length, total_duration
+        sub_dataframe_shape, tracklet_length, total_duration
     )
     sub_arrays_dynamic = sample_subarrays(
-        sub_dataframe_dynamic, num_samples, tracklet_length, total_duration
+        sub_dataframe_dynamic, tracklet_length, total_duration
     )
 
     def make_prediction(input_data, model):
@@ -4089,12 +4092,11 @@ def make_prediction(input_data, model):
         return predicted_class.item()
 
     def get_most_frequent_prediction(predictions):
-        if predictions:
 
-            prediction_counts = Counter(predictions)
-            most_common_prediction, count = prediction_counts.most_common(1)[0]
+        prediction_counts = Counter(predictions)
+        most_common_prediction, count = prediction_counts.most_common(1)[0]
 
-            return most_common_prediction
+        return most_common_prediction
 
     shape_predictions = []
     if shape_model is not None:
@@ -4109,16 +4111,11 @@ def get_most_frequent_prediction(predictions):
             dynamic_predictions.append(predicted_class)
 
     final_predictions = shape_predictions + dynamic_predictions
-    if len(final_predictions) > 0:
-        most_frequent_prediction = get_most_frequent_prediction(final_predictions)
-        if most_frequent_prediction is not None:
-            most_predicted_class = class_map[int(most_frequent_prediction)]
-
-            return most_predicted_class
+    most_frequent_prediction = get_most_frequent_prediction(final_predictions)
 
-    else:
+    most_predicted_class = class_map[int(most_frequent_prediction)]
 
-        return "UnClassified"
+    return most_predicted_class
 
 
 def save_cell_type_predictions(

src/napatrackmater/_version.py

@@ -1,2 +1,2 @@
-__version__ = version = "5.4.9"
-__version_tuple__ = version_tuple = (5, 4, 9)
+__version__ = version = "5.5.0"
+__version_tuple__ = version_tuple = (5, 5, 0)