.gitignore

-*.jpg
+
 .DS_Store
 
 # Data

utils/compare_models_accuracy.py

+import numpy as np
+
+def compare_models_accuracy(classes, class_correct_1, class_correct_2, class_total, model_name_1, model_name_2):
+    # Compare les résultats de deux modèles (accuracy par classe)
+    print("Overall Accuracy:")   
+    print(
+            "%2s: %2d%% (%2d/%2d) \t %s: %2d%% (%2d/%2d) --> %s%4s%%"
+            % (
+                model_name_1,
+                100.0 * np.sum(class_correct_1) / np.sum(class_total),
+                np.sum(class_correct_1),
+                np.sum(class_total),
+                model_name_2,
+                100.0 * np.sum(class_correct_2) / np.sum(class_total),
+                np.sum(class_correct_2),
+                np.sum(class_total),
+                "+" if np.sum(class_correct_2) - np.sum(class_correct_1) > 0 else "",
+                100.0 * (np.sum(class_correct_2) - np.sum(class_correct_1)) / np.sum(class_total)
+            )
+        )
+    print("\nAccuracy by class:")
+    for i in range(10):
+        if class_total[i] > 0:
+            print(
+                "%10s \t %2s: %2d%% (%2d/%2d) \t %s: %2d%% (%2d/%2d) --> %s%4s%%"
+                % (
+                    classes[i],
+                    model_name_1,
+                    100 * class_correct_1[i] / class_total[i],
+                    np.sum(class_correct_1[i]),
+                    np.sum(class_total[i]),
+                    model_name_2,
+                    100 * class_correct_2[i] / class_total[i],
+                    np.sum(class_correct_2[i]),
+                    np.sum(class_total[i]),
+                    "+" if class_correct_2[i] - class_correct_1[i] > 0 else "",
+                    100 * (class_correct_2[i] - class_correct_1[i]) / class_total[i]
+                )
+            )
\ No newline at end of file

utils/plot_performance.py

+import matplotlib.pyplot as plt
+
+
+def plot_performance(train_loss_list, val_loss_list, title = "Performance of Model", save_path = None):
+    # Plot the performance of the model (train loss and validation loss)
+    epochs_list = range(len(train_loss_list))
+    plt.plot(epochs_list, train_loss_list)
+    plt.plot(epochs_list, val_loss_list)
+    plt.legend(["Train Loss", "Validation Loss"])
+    plt.xlabel("Epoch")
+    plt.ylabel("Loss")
+    plt.title(title)
+    if save_path is not None:
+        plt.savefig(save_path)
+    plt.show()
\ No newline at end of file

utils/print_size_of_model.py

+import os
+import torch
+
+def print_size_of_model(model, label=""):
+    # Print the size of the model
+    torch.save(model.state_dict(), "temp.p")
+    size = os.path.getsize("temp.p")
+    print("model: ", label, " \t", "Size (KB):", size / 1e3)
+    os.remove("temp.p")
+    return size
\ No newline at end of file