-
Notifications
You must be signed in to change notification settings - Fork 0
/
coding_test_iisc.py
268 lines (209 loc) · 9.33 KB
/
coding_test_iisc.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
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
264
265
266
267
# -*- coding: utf-8 -*-
"""Coding_test_IISc.ipynb
Automatically generated by Colaboratory.
Original file is located at
https://colab.research.google.com/drive/104tvSdBDm2Th1fMHZrpfmVswAANRM5wF
"""
# importing the required libraries
from __future__ import print_function
from __future__ import division
import torch
import torch.nn as nn
import torch.optim as optim
import numpy as np
import torchvision
from torch.utils.data import Subset, DataLoader
from torchvision import datasets, models, transforms
import torch.nn.functional as F
import matplotlib.pyplot as plt
import time
import os
import copy
import sys
import os
import logging
from PIL import Image
from tqdm import tqdm
# !unzip drive/MyDrive/PACS.zip &> /dev/null
# from PIL import Image
# Image.open('/content/kfold/cartoon/dog/pic_016.jpg').size
def set_parameter_requires_grad(model, feature_extracting):
if feature_extracting:
for param in model.parameters():
param.requires_grad = False
def initialize_model(num_classes, feature_extract = True, use_pretrained=True):
model_ft = None
input_size = 0
model_ft = models.resnet18(pretrained=use_pretrained) #Using pretrained feature extractor
set_parameter_requires_grad(model_ft, feature_extract) #Setting .reuires_grad to False so that params remains fixed for feature extractor
num_ftrs = model_ft.fc.in_features
model_ft.fc = nn.Linear(num_ftrs, num_classes) #Classifier
return model_ft
feature_extract = True
model_ft = initialize_model( 7, feature_extract = True, use_pretrained=True) #for determining object class
lamda_param = torch.Tensor([0.43,0.23, 0.33]) # randomly initialized
# taking only those parameters that are to be optimized
if feature_extract:
params_to_update = []
for name,param in model_ft.named_parameters():
if param.requires_grad == True:
params_to_update.append(param)
print("\t",name)
params_to_update.append(lamda_param)
optimizer_ft = optim.SGD(params_to_update, lr=0.001, momentum=0.9)
print(model_ft)
if not os.path.isdir('./Homework3-PACS'):
!git clone https://github.com/MachineLearning2020/Homework3-PACS
# Define datasets root
DIR_PHOTO = 'Homework3-PACS/PACS/photo'
DIR_ART = 'Homework3-PACS/PACS/art_painting'
DIR_CARTOON = 'Homework3-PACS/PACS/cartoon'
DIR_SKETCH = 'Homework3-PACS/PACS/sketch'
# means and standard deviations ImageNet because the network is pretrained
means, stds = (0.485, 0.456, 0.406), (0.229, 0.224, 0.225)
transf = transforms.Compose([transforms.CenterCrop(224), # Crops a central square patch of the image 224 because torchvision's AlexNet needs a 224x224 input!
transforms.ToTensor(), # Turn PIL Image to torch.Tensor
transforms.Normalize(means,stds) # Normalizes tensor with mean and standard deviation
])
# Prepare Pytorch train/test Datasets
photo_dataset = torchvision.datasets.ImageFolder(DIR_PHOTO, transform=transf)
art_dataset = torchvision.datasets.ImageFolder(DIR_ART, transform=transf)
cartoon_dataset = torchvision.datasets.ImageFolder(DIR_CARTOON, transform=transf)
sketch_dataset = torchvision.datasets.ImageFolder(DIR_SKETCH, transform=transf)
# Check dataset sizes
print(f"Photo Dataset: {len(photo_dataset)}")
print(f"Art Dataset: {len(art_dataset)}")
print(f"Cartoon Dataset: {len(cartoon_dataset)}")
print(f"Sketch Dataset: {len(sketch_dataset)}")
BATCH_SIZE = 128
# Dataloaders iterate over pytorch datasets
photo_dataloader = DataLoader(photo_dataset, batch_size=BATCH_SIZE, shuffle=True, num_workers=2)
art_dataloader = DataLoader(art_dataset, batch_size=BATCH_SIZE, shuffle=True, num_workers=2, drop_last=False)
cartoon_dataloader = DataLoader(cartoon_dataset, batch_size=BATCH_SIZE, shuffle=True, num_workers=2, drop_last=False)
sketch_dataloader = DataLoader(sketch_dataset, batch_size=BATCH_SIZE, shuffle=True, num_workers=2, drop_last=False)
from sklearn.model_selection import train_test_split
def train_val_dataset(dataset, val_split=0.1):
train_idx, val_idx = train_test_split(list(range(len(dataset))), test_size=val_split)
datasets = {}
train_source_data = Subset(dataset, train_idx)
val_source_data = Subset(dataset, val_idx)
return train_source_data, val_source_data
l = []
l.append(photo_dataset)
l.append(art_dataset)
l.append(cartoon_dataset)
image_datasets = torch.utils.data.ConcatDataset(l)
train_source_data, val_source_data = train_val_dataset(image_datasets, val_split = 0.1)
#prepare train and val dataloader
train_dataloader = DataLoader(train_source_data, batch_size=BATCH_SIZE, shuffle=True, num_workers=2)
val_dataloader = DataLoader(val_source_data, batch_size=BATCH_SIZE, shuffle=True, num_workers=2)
test_loader = sketch_dataloader
def LossCE():
return nn.CrossEntropyLoss()
def LossWeighted(lossce, sigma):
sig_max = np.max(sigma)
if sig_max>= 0.95:
return sig_max * lossce
else:
return 0
def Hloss():
return nn.CrossEntropyLoss()
dataloaders = {'train': train_dataloader, 'val': val_dataloader}
device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu")
loss1_list = []
loss2_list = []
loss3_list = []
def train_model(model, dataloaders, criterion, optimizer, num_epochs=25, is_inception=False):
since = time.time()
val_acc_history = []
best_model_wts = copy.deepcopy(model.state_dict())
best_acc = 0.0
for epoch in range(num_epochs):
print('Epoch {}/{}'.format(epoch, num_epochs - 1))
print('-' * 10)
# Each epoch has a training and validation phase
for phase in ['train', 'val']:
if phase == 'train':
model.train() # Set model to training mode
else:
model.eval() # Set model to evaluate mode
running_loss = 0.0
running_corrects = 0
# Iterate over data.
for inputs, labels in dataloaders[phase]:
inputs = inputs.to(device)
labels = labels.to(device)
# zero the parameter gradients
optimizer.zero_grad()
# forward
# track history if only in train
with torch.set_grad_enabled(phase == 'train'):
if is_inception and phase == 'train':
outputs = model(inputs)
sigma = F.softmax(outputs, dim=1)
sig_max = torch.max(sigma)
if sig_max >= 0.95:
loss2 = sig_max*criterion(outputs, labels)
else:
loss2 = 0
loss2_list.append(loss2)
loss1 = criterion(outputs, labels)
loss1_list.append(loss1)
loss3 = criterion(outputs, labels)
loss3_list.append(loss3)
loss = lamda_param[0]*loss1+lamda_param[1]*loss2+lamda_param[2]*loss3
else:
outputs = model(inputs)
loss = criterion(outputs, labels)
_, preds = torch.max(outputs, 1)
# backward + optimize only if in training phase
if phase == 'train':
loss.backward()
optimizer.step()
# statistics
running_loss += loss.item() * inputs.size(0)
running_corrects += torch.sum(preds == labels.data)
epoch_loss = running_loss / len(dataloaders[phase].dataset)
epoch_acc = running_corrects.double() / len(dataloaders[phase].dataset)
print('{} Loss: {:.4f} Acc: {:.4f}'.format(phase, epoch_loss, epoch_acc))
# deep copy the model
if phase == 'val' and epoch_acc > best_acc:
best_acc = epoch_acc
best_model_wts = copy.deepcopy(model.state_dict())
if phase == 'val':
val_acc_history.append(epoch_acc)
print()
time_elapsed = time.time() - since
print('Training complete in {:.0f}m {:.0f}s'.format(time_elapsed // 60, time_elapsed % 60))
print('Best val Acc: {:4f}'.format(best_acc))
# load best model weights
model.load_state_dict(best_model_wts)
return model, val_acc_history
model_ft, hist = train_model(model_ft, dataloaders, criterion = nn.CrossEntropyLoss(), optimizer = optimizer_ft, num_epochs=2, is_inception=True)
# to make a graph tracking the status of individual loss components as the training progresses
def plotLosses( loss_2, loss_3, n_epochs=10, save_pic=True) :
epochs = range(n_epochs)
plt.figure()
plt.xlabel('Iterations')
plt.ylabel('Loss')
plt.plot(epochs, loss_2, 'g--', label="LossWeighted")
plt.plot(epochs, loss_3, 'r--', label="HLoss")
plt.legend()
plt.show()
if save_pic:
plt.savefig('losses.png')
return
plotLosses( loss2_list, loss3_list, n_epochs=len(loss1_list), save_pic = True)
# finding the test accuracy
running_corrects = 0
model_ft.eval()
acc_pro = []
for inputs, labels in test_loader:
inputs = inputs.to(device)
labels = labels.to(device)
outputs = model_ft(inputs)
_, preds = torch.max(outputs, 1)
running_corrects += torch.sum(preds == labels.data)
epoch_acc = running_corrects.double() / len(test_loader.dataset)
acc_pro.append(epoch_acc)
print('Test Accuracy: {:.4f}'.format(acc_pro[-1]))