-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathmain.py
181 lines (157 loc) · 9.03 KB
/
main.py
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
import os
import argparse
import random
import warnings
from sklearn.feature_selection import mutual_info_classif
from src.utils.image_processing import image_handler
from src.pipelines import train_pipeline, predict_pipeline
from src.components import data_ingestion, data_transformation, model_training, model_prediction
from models.test_models import models, models_small
from src.logger import logging
def parse_args():
parser = argparse.ArgumentParser(description='Train or predict with model')
parser.add_argument('--train', default=False, action='store_true', help='Train model')
parser.add_argument('--predict', default=False, action='store_true', help='Use model')
parser.add_argument('--id_column', type=str, default='image_id', help='ID column name')
parser.add_argument('--target_column', type=str, default='label', help='Target column name')
parser.add_argument('--positive_class', type=str, default='pools', help='Positive class name')
parser.add_argument('--negative_class', type=str, default='no_pools', help='Negative class name')
parser.add_argument('--cache_features', default=False, action='store_true', help='Cache features')
parser.add_argument('--seed', type=int, default=0, help='Random seed')
# Train pipeline arguments
parser.add_argument('--train_images_path', type=str, default=os.path.join('data', 'datasets', 'yacine', 'custom_splits', 'train'), help='Path to train images')
parser.add_argument('--val_images_path', type=str, default=os.path.join('data', 'datasets', 'yacine', 'custom_splits', 'validation'), help='Path to validate images')
parser.add_argument('--test_images_path', type=str, default=os.path.join('data', 'datasets', 'yacine', 'original_splits', 'validation'), help='Path to test images')
parser.add_argument('--train_model_save_path', type=str, default=os.path.join('models', 'best_model.pkl'), help='Path to save model')
parser.add_argument('--train_results_save_path', type=str, default=os.path.join('results', 'results.csv'), help='Path to save results')
parser.add_argument('--train_features_save_path', type=str, default=os.path.join('data', 'train_features.csv'), help='Path to save train features')
parser.add_argument('--validation_features_save_path', type=str, default=os.path.join('data', 'validation_features.csv'), help='Path to save validation features')
parser.add_argument('--test_features_save_path', type=str, default=os.path.join('data', 'test_features.csv'), help='Path to save test features')
parser.add_argument('--score_criteria', type=str, default='accuracy', help='Score criteria to select best model')
parser.add_argument('--small_grid', default=False, action='store_true', help='Use small grid for training')
parser.add_argument('--not_drop_correlated_features', default=False, action='store_true', help='Does not drop correlated features when activated')
parser.add_argument('--correlated_features_path', type=str, default=os.path.join('data', 'features', 'correlated_features.txt'), help='Path to save correlated features')
# Predict pipeline arguments
parser.add_argument('--k_features', type=int, default=50, help='Number of features to select')
parser.add_argument('--use_gabor', type=int, default=0, help='Use gabor filter')
parser.add_argument('--predict_data_path', type=str, default=os.path.join('data', 'datasets', 'algarves', 'fragmented_dataset'), help='Path to predict images')
parser.add_argument('--predict_model_path', type=str, default=os.path.join('models', 'best_model.pkl'), help='Path to save model')
parser.add_argument('--predict_features_save_path', type=str, default=os.path.join('data', 'predict_features.csv'), help='Path to save predict features')
parser.add_argument('--predict_results_save_path', type=str, default=os.path.join('results', 'prediction_results.csv'), help='Path to save results')
return vars(parser.parse_args())
def main(args):
seed = args['seed']
random.seed(seed)
warnings.filterwarnings('ignore')
color_features = ['has_blue']
channel_features = ['mean',
'std',
'median',
'mode',
'min',
'max',
'range',
'skewness',
'kurtosis',
'entropy',
'quantile_0.25',
'quantile_0.75',
'iqr']
histogram_features = ['mean',
'std',
'median',
'mode',
'min',
'max',
'range',
'skewness',
'kurtosis',
'uniformity',
'entropy',
'R']
coocurrence_matrix_features = ['contrast',
'dissimilarity',
'homogeneity',
'energy',
'correlation']
with open(args['correlated_features_path'], 'r') as f:
correlated_features = [t.strip() for t in f.readlines()]
k_features_grid = [20, 30, 40, 'all']
use_gabor_grid = [0]
use_augmentation_grid = [0, 1]
# Data Ingestion
ingestion_config = data_ingestion.DataIngestionConfig(
train_data_path=args['train_images_path'],
val_data_path=args['val_images_path'],
test_data_path=args['test_images_path'],
predict_data_path=args['predict_data_path'],
load_images=image_handler.load_image)
data_ingestor = data_ingestion.DataIngestor(ingestion_config)
# Data Transformation
transformation_config = data_transformation.DataTransformationConfig(
color_features=color_features,
channel_features=channel_features,
histogram_features=histogram_features,
coocurrence_matrix_features=coocurrence_matrix_features,
correlated_features=correlated_features,
drop_correlated_features=not args['not_drop_correlated_features'],
use_augmentation=1 if args['train'] or 1 in use_augmentation_grid else 0,
positive_class=args['positive_class'],
negative_class=args['negative_class'],
to_grayscale=image_handler.to_grayscale,
to_histogram=image_handler.to_histogram)
data_transformer = data_transformation.DataTransformer(transformation_config)
# Model Training/Prediction
if args['train']:
train_models = models_small if args['small_grid'] else models
model_config = model_training.ModelTrainerConfig(
k_features_grid=k_features_grid,
use_gabor_grid=use_gabor_grid,
use_augmentation_grid=use_augmentation_grid,
feature_selection_score_function=mutual_info_classif,
models=train_models,
id_column=args['id_column'],
target_column=args['target_column'],
score_criteria=args['score_criteria'])
model_trainer = model_training.ModelTrainer(model_config)
pipeline_config = train_pipeline.TrainPipelineConfig(
model_save_path=args['train_model_save_path'],
results_save_path=args['train_results_save_path'],
train_features_save_path=args['train_features_save_path'],
validation_features_save_path=args['validation_features_save_path'],
test_features_save_path=args['test_features_save_path'],
cache_features=args['cache_features'],
logger=logging)
pipeline = train_pipeline.TrainPipeline(
config=pipeline_config,
data_ingestor=data_ingestor,
data_transformer=data_transformer,
model_trainer=model_trainer)
pipeline.train()
elif args['predict']:
model_config = model_prediction.ModelPredictorConfig(
k_features=args['k_features'],
use_gabor=args['use_gabor'],
feature_selection_score_function=mutual_info_classif,
id_column=args['id_column'],
target_column=args['target_column'])
model_predictor = model_prediction.ModelPredictor(model_config)
pipeline_config = predict_pipeline.PredictPipelineConfig(
model_path=args['predict_model_path'],
results_save_path=args['predict_results_save_path'],
features_save_path=args['predict_features_save_path'],
cache_features=args['cache_features'],
logger=logging)
pipeline = predict_pipeline.PredictPipeline(
config=pipeline_config,
data_ingestor=data_ingestor,
data_transformer=data_transformer,
model_predictor=model_predictor)
pipeline.predict()
else:
print('No action selected')
# python main.py --train --small_grid --cache_features
# python main.py --train --small_grid --train_images_path data/datasets/mix/train
if __name__ == '__main__':
args = parse_args()
main(args)