Object-Verification/demo_train_model.py at main · MaxCoppa/Object-Verification · 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
# %% -------------------------
# Import necessary libraries
# -------------------------
import time
import torch
import numpy as np
import os

from torch.utils.data import DataLoader
import torch.optim as optim

# Project modules
from dataset_annotation_preparation import prepare_annotation
from dataset_preloader import VeriImageDataset
from utils import image_utils, model_utils
from object_verif_models import ObjectVeriSiamese
from training_evaluation import train_object_verif_model

# Import project configuration
from config import (
    ANNOTATIONS_PATH,
    RAW_ANNOTATIONS_PATH,
    DATA_PATH,
    MODELS_PATH,
    LOGS_PATH,
)

# %% -------------------------
# Define parameters
# -------------------------
train_ratio = 0.5
n_error = 2
n_augmentation = 100

load = True
transform_type = "test"

raw_annotation_sharks = RAW_ANNOTATIONS_PATH / "raw_annotations_sharks.csv"
preprocessed_annotation_sharks = ANNOTATIONS_PATH / "train_annotations_sharks.csv"
images_dir = DATA_PATH / "animals"

# %% -------------------------
# Test Data Annotation
# -------------------------
df = prepare_annotation(
    raw_annotation_path=raw_annotation_sharks,
    images_dir=images_dir,
    preprocessed_annotation_path=preprocessed_annotation_sharks,
    train_ratio=train_ratio,
    n_augmentation=n_augmentation,
    n_error=n_error,
)

# %% -------------------------
# Training setup
# -------------------------
frozen = True
loss_name = "Contrastiveloss"
model_name = "2"
transform_type_train = "transform_data_aug"
transform_type_val = "test"

n_augmentation = 1
backbone = "mobilenet_v3_small"
batch_size = 64
num_epochs = 1

# Loss function
criterion = model_utils.get_loss_function(loss_name)

# Data transforms
transform_train = image_utils.transform_fc(transform_type_train)
transform_val = image_utils.transform_fc(transform_type_val)

# Datasets
train_data = VeriImageDataset(
    annotations_file=preprocessed_annotation_sharks,
    train=True,
    transform=transform_train,
    crop_type=None,
)

val_data = VeriImageDataset(
    annotations_file=preprocessed_annotation_sharks,
    train=False,
    transform=transform_val,
    crop_type=None,
)

print(len(train_data), len(val_data))

# %% -------------------------
# Dataloaders
# -------------------------
dataloaders = {
    "train": DataLoader(
        train_data,
        batch_size=batch_size,
        shuffle=True,
        num_workers=0,
        pin_memory=True,
    ),
    "val": DataLoader(
        val_data,
        batch_size=batch_size,
        shuffle=True,
        num_workers=0,
        pin_memory=True,
    ),
}

# %% -------------------------
# Model setup
# -------------------------
model = ObjectVeriSiamese(backbone=backbone, freeze_backbone=frozen)
if frozen:
    params = model.fc.parameters()
else:
    params = model.parameters()

optimizer = optim.SGD(params, lr=0.01, momentum=0.9, weight_decay=0.0001)

# %% -------------------------
# Train model
# -------------------------
model = train_object_verif_model(
    model,
    criterion=criterion,
    optimizer=optimizer,
    dataloaders=dataloaders,
    num_epochs=num_epochs,
    freeze_backbone=frozen,
    save_path=MODELS_PATH / f"model_{model_name}.pth",
    log_filename=LOGS_PATH / f"log_{model_name}.log",
    log_to_console=True,
    verbose=True,
)

# %%