Classification-instruments/main.py at main · GiovanniIacuzzo/Classification-instruments · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
import torch
import torch.optim as optim
import torch.nn as nn
from utils.train_immagini import train_model
from utils.test_immagini import test_model
from utils.dataset_immagini import ImmaginiDataset
from models.model_immagini import CNNModel

if __name__ == "__main__":
    batch_size = 64
    img_size = 224
    num_epochs = 50
    learning_rate = 0.001
    patience = 10
    device = torch.device('mps' if torch.cuda.is_available() else 'cpu')

    torch.manual_seed(42)

    train_dataset = ImmaginiDataset(root_dir='./dataset_diviso', subset='train', img_size=224)
    val_dataset   = ImmaginiDataset(root_dir='./dataset_diviso', subset='val', img_size=224)
    test_dataset  = ImmaginiDataset(root_dir='./dataset_diviso', subset='test', img_size=224)

    train_loader = torch.utils.data.DataLoader(train_dataset, batch_size=32, shuffle=True)
    val_loader   = torch.utils.data.DataLoader(val_dataset, batch_size=32, shuffle=False)
    test_loader  = torch.utils.data.DataLoader(test_dataset, batch_size=32, shuffle=False)


    model = CNNModel(num_classes=len(train_dataset.classes)).to(device)
    criterion = nn.CrossEntropyLoss()
    optimizer = optim.Adam(model.parameters(), lr=learning_rate)

    train_model(train_loader, val_loader, model, criterion, optimizer, num_epochs=num_epochs, device=device, patience=patience)

    print("\nCaricamento del modello migliore...")
    model.load_state_dict(torch.load('best_model.pth'))

    print("\nEsecuzione del test sul dataset di test...")
    test_dataset.classes = [c for c in test_dataset.classes if not c.startswith('.')]
    test_model(test_loader, model, device=device, class_names=test_dataset.classes)


"""
(base) giovanni02@MacBook-Air-del-Professore Classification-instruments % /usr/local/bin/python3 /Users/giovanni02/Desktop/Progetti/Classifi
cation-instruments/main.py

Epoch [1/50]
Train Loss: 15.8326, Train Acc: 58.35%
Val Loss: 10.9361, Val Acc: 54.91%

Epoch [2/50]
Train Loss: 3.6461, Train Acc: 77.35%
Val Loss: 3.8957, Val Acc: 73.41%

Epoch [3/50]
Train Loss: 1.5649, Train Acc: 85.28%
Val Loss: 4.1540, Val Acc: 76.30%

Epoch [4/50]
Train Loss: 1.0558, Train Acc: 86.33%
Val Loss: 1.5769, Val Acc: 76.88%

Epoch [5/50]
Train Loss: 0.3686, Train Acc: 87.58%
Val Loss: 1.6762, Val Acc: 80.35%

Epoch [6/50]
Train Loss: 0.3186, Train Acc: 88.62%
Val Loss: 1.7355, Val Acc: 82.08%

Epoch [7/50]
Train Loss: 0.4412, Train Acc: 86.33%
Val Loss: 1.4512, Val Acc: 79.19%

Epoch [8/50]
Train Loss: 0.3029, Train Acc: 88.83%
Val Loss: 1.6125, Val Acc: 78.61%

Epoch [9/50]
Train Loss: 0.2798, Train Acc: 88.20%
Val Loss: 1.9081, Val Acc: 78.61%

Epoch [10/50]
Train Loss: 0.2652, Train Acc: 90.92%
Val Loss: 1.8789, Val Acc: 82.08%

Epoch [11/50]
Train Loss: 0.3205, Train Acc: 89.14%
Val Loss: 1.6639, Val Acc: 83.24%

Epoch [12/50]
Train Loss: 0.2560, Train Acc: 89.98%
Val Loss: 1.7430, Val Acc: 83.24%

Epoch [13/50]
Train Loss: 0.3312, Train Acc: 87.79%
Val Loss: 1.5559, Val Acc: 81.50%

Epoch [14/50]
Train Loss: 0.2730, Train Acc: 90.19%
Val Loss: 1.5545, Val Acc: 83.24%

Epoch [15/50]
Train Loss: 0.2179, Train Acc: 91.23%
Val Loss: 1.4229, Val Acc: 84.97%

Epoch [16/50]
Train Loss: 0.2298, Train Acc: 91.34%
Val Loss: 1.2727, Val Acc: 83.24%

Epoch [17/50]
Train Loss: 0.2056, Train Acc: 92.90%
Val Loss: 1.3538, Val Acc: 84.97%

Epoch [18/50]
Train Loss: 0.1862, Train Acc: 94.15%
Val Loss: 1.7075, Val Acc: 83.82%

Epoch [19/50]
Train Loss: 0.2193, Train Acc: 92.48%
Val Loss: 1.3306, Val Acc: 84.39%

Epoch [20/50]
Train Loss: 0.1530, Train Acc: 93.95%
Val Loss: 1.7196, Val Acc: 80.92%

Epoch [21/50]
Train Loss: 0.2277, Train Acc: 91.13%
Val Loss: 1.2769, Val Acc: 83.82%

Epoch [22/50]
Train Loss: 0.1936, Train Acc: 92.69%
Val Loss: 1.4976, Val Acc: 84.39%

Epoch [23/50]
Train Loss: 0.2781, Train Acc: 90.40%
Val Loss: 1.3629, Val Acc: 83.24%

Epoch [24/50]
Train Loss: 0.1965, Train Acc: 92.48%
Val Loss: 1.5408, Val Acc: 84.39%

Epoch [25/50]
Train Loss: 0.1584, Train Acc: 95.20%
Val Loss: 1.0970, Val Acc: 85.55%

Epoch [26/50]
Train Loss: 0.2252, Train Acc: 91.86%
Val Loss: 1.5446, Val Acc: 82.66%

Epoch [27/50]
Train Loss: 0.1932, Train Acc: 91.75%
Val Loss: 1.8867, Val Acc: 86.13%

Epoch [28/50]
Train Loss: 0.1868, Train Acc: 93.74%
Val Loss: 1.6581, Val Acc: 85.55%

Epoch [29/50]
Train Loss: 0.1539, Train Acc: 94.47%
Val Loss: 1.2282, Val Acc: 85.55%

Epoch [30/50]
Train Loss: 0.1397, Train Acc: 94.78%
Val Loss: 1.5233, Val Acc: 83.82%

Epoch [31/50]
Train Loss: 0.1009, Train Acc: 95.51%
Val Loss: 2.2278, Val Acc: 86.71%

Epoch [32/50]
Train Loss: 0.1156, Train Acc: 95.72%
Val Loss: 1.7122, Val Acc: 88.44%

Epoch [33/50]
Train Loss: 0.0984, Train Acc: 96.35%
Val Loss: 1.7169, Val Acc: 87.28%

Epoch [34/50]
Train Loss: 0.1154, Train Acc: 95.20%
Val Loss: 1.2143, Val Acc: 87.86%

Epoch [35/50]
Train Loss: 0.0999, Train Acc: 96.56%
Val Loss: 1.2161, Val Acc: 89.02%

Epoch [36/50]
Train Loss: 0.1028, Train Acc: 95.72%
Val Loss: 0.8458, Val Acc: 87.86%

Epoch [37/50]
Train Loss: 0.0904, Train Acc: 96.66%
Val Loss: 1.1310, Val Acc: 88.44%

Epoch [38/50]
Train Loss: 0.0926, Train Acc: 96.87%
Val Loss: 1.4538, Val Acc: 88.44%

Epoch [39/50]
Train Loss: 0.0935, Train Acc: 96.35%
Val Loss: 1.3201, Val Acc: 88.44%

Epoch [40/50]
Train Loss: 0.1119, Train Acc: 96.03%
Val Loss: 0.8750, Val Acc: 87.86%

Epoch [41/50]
Train Loss: 0.1273, Train Acc: 95.62%
Val Loss: 0.8118, Val Acc: 89.02%

Epoch [42/50]
Train Loss: 0.0665, Train Acc: 97.08%
Val Loss: 1.2738, Val Acc: 88.44%

Epoch [43/50]
Train Loss: 0.0682, Train Acc: 97.18%
Val Loss: 0.9687, Val Acc: 87.86%

Epoch [44/50]
Train Loss: 0.1076, Train Acc: 96.03%
Val Loss: 1.3153, Val Acc: 87.28%

Epoch [45/50]
Train Loss: 0.0948, Train Acc: 96.56%
Val Loss: 1.6787, Val Acc: 86.71%
Early stopping triggered

Best Validation Accuracy: 89.02%

Final Evaluation Report:

              precision    recall  f1-score   support

           0     0.6727    1.0000    0.8043        37
           1     1.0000    0.9730    0.9863        37
           2     1.0000    0.9545    0.9767        22
           3     0.9024    0.9737    0.9367        38
           4     0.9500    0.4872    0.6441        39

    accuracy                         0.8671       173
   macro avg     0.9050    0.8777    0.8696       173
weighted avg     0.8973    0.8671    0.8581       173


Caricamento del modello migliore...

Esecuzione del test sul dataset di test...

TEST REPORT:

              precision    recall  f1-score   support

    chitarra     1.0000    1.0000    1.0000        30
      flauto     1.0000    0.9767    0.9882        43
  pianoforte     0.9655    1.0000    0.9825        28
       viola     0.9744    0.9744    0.9744        39
     violino     0.9750    0.9750    0.9750        40

    accuracy                         0.9833       180
   macro avg     0.9830    0.9852    0.9840       180
weighted avg     0.9835    0.9833    0.9833       180

(base) giovanni02@MacBook-Air-del-Professore Classification-instruments %
"""