File size: 10,380 Bytes
a6eed2b |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 |
import torch
import torch.nn as nn
import torch.nn.functional as F
import numpy as np
from torchvision import transforms
import random
def cutmix_data(x, y, alpha=1.0, device='cuda'):
"""
CutMix data augmentation.
Args:
x: Input batch
y: Target batch
alpha: CutMix parameter
device: Device to run on
Returns:
mixed_x: Mixed input batch
y_a, y_b: Original targets for loss calculation
lam: Mixing ratio
"""
if alpha > 0:
lam = np.random.beta(alpha, alpha)
else:
lam = 1
batch_size = x.size(0)
if device == 'cuda':
index = torch.randperm(batch_size).cuda()
else:
index = torch.randperm(batch_size)
# Generate random bounding box
W = x.size(2)
H = x.size(3)
cut_rat = np.sqrt(1. - lam)
cut_w = int(W * cut_rat)
cut_h = int(H * cut_rat)
# Uniform sampling
cx = np.random.randint(W)
cy = np.random.randint(H)
bbx1 = np.clip(cx - cut_w // 2, 0, W)
bby1 = np.clip(cy - cut_h // 2, 0, H)
bbx2 = np.clip(cx + cut_w // 2, 0, W)
bby2 = np.clip(cy + cut_h // 2, 0, H)
x[:, :, bbx1:bbx2, bby1:bby2] = x[index, :, bbx1:bbx2, bby1:bby2]
# Adjust lambda to exactly match pixel ratio
lam = 1 - ((bbx2 - bbx1) * (bby2 - bby1) / (W * H))
y_a, y_b = y, y[index]
return x, y_a, y_b, lam
def cutmix_criterion(criterion, pred, y_a, y_b, lam):
"""
CutMix loss calculation.
"""
return lam * criterion(pred, y_a) + (1 - lam) * criterion(pred, y_b)
class LabelSmoothingCrossEntropy(nn.Module):
"""
Label smoothing cross entropy loss.
"""
def __init__(self, smoothing=0.1):
super(LabelSmoothingCrossEntropy, self).__init__()
self.smoothing = smoothing
def forward(self, x, target):
confidence = 1. - self.smoothing
logprobs = F.log_softmax(x, dim=-1)
nll_loss = -logprobs.gather(dim=-1, index=target.unsqueeze(1))
nll_loss = nll_loss.squeeze(1)
smooth_loss = -logprobs.mean(dim=-1)
loss = confidence * nll_loss + self.smoothing * smooth_loss
return loss.mean()
class FocalLoss(nn.Module):
"""
Focal Loss for addressing class imbalance.
"""
def __init__(self, alpha=1, gamma=2, reduction='mean'):
super(FocalLoss, self).__init__()
self.alpha = alpha
self.gamma = gamma
self.reduction = reduction
def forward(self, inputs, targets):
ce_loss = F.cross_entropy(inputs, targets, reduction='none')
pt = torch.exp(-ce_loss)
focal_loss = self.alpha * (1-pt)**self.gamma * ce_loss
if self.reduction == 'mean':
return focal_loss.mean()
elif self.reduction == 'sum':
return focal_loss.sum()
else:
return focal_loss
class AdvancedAugmentation:
"""
Advanced augmentation techniques for better generalization.
"""
def __init__(self, image_size=224):
self.image_size = image_size
def get_train_transforms(self):
"""
Get comprehensive training transforms with advanced augmentation.
"""
return transforms.Compose([
# Resize with padding
transforms.Resize((self.image_size + 32, self.image_size + 32)),
# Random crop with padding
transforms.RandomCrop((self.image_size, self.image_size), padding=4),
# Geometric augmentations
transforms.RandomHorizontalFlip(p=0.5),
transforms.RandomVerticalFlip(p=0.2),
transforms.RandomRotation(degrees=15),
transforms.RandomAffine(
degrees=0,
translate=(0.1, 0.1),
scale=(0.9, 1.1),
shear=5
),
# Color augmentations
transforms.ColorJitter(
brightness=0.2,
contrast=0.2,
saturation=0.2,
hue=0.05
),
# Advanced augmentations
transforms.RandomPerspective(distortion_scale=0.2, p=0.3),
transforms.RandomErasing(p=0.2, scale=(0.02, 0.33), ratio=(0.3, 3.3)),
# TrivialAugmentWide for additional randomness
transforms.TrivialAugmentWide(num_magnitude_bins=31),
# Convert to tensor and normalize
transforms.ToTensor(),
transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
])
def get_val_transforms(self):
"""
Get validation transforms (minimal augmentation).
"""
return transforms.Compose([
transforms.Resize((self.image_size, self.image_size)),
transforms.ToTensor(),
transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
])
class TestTimeAugmentation:
"""
Test Time Augmentation for better inference.
"""
def __init__(self, model, device, num_augmentations=5):
self.model = model
self.device = device
self.num_augmentations = num_augmentations
# Define TTA transforms
self.tta_transforms = [
transforms.Compose([
transforms.Resize((224, 224)),
transforms.ToTensor(),
transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
]),
transforms.Compose([
transforms.Resize((224, 224)),
transforms.RandomHorizontalFlip(p=1.0),
transforms.ToTensor(),
transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
]),
transforms.Compose([
transforms.Resize((224, 224)),
transforms.RandomRotation(degrees=10),
transforms.ToTensor(),
transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
]),
transforms.Compose([
transforms.Resize((224, 224)),
transforms.RandomRotation(degrees=10),
transforms.ToTensor(),
transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
]),
transforms.Compose([
transforms.Resize((224, 224)),
transforms.ColorJitter(brightness=0.1, contrast=0.1),
transforms.ToTensor(),
transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
])
]
def predict(self, image):
"""
Predict with TTA.
"""
self.model.eval()
predictions = []
with torch.no_grad():
for transform in self.tta_transforms[:self.num_augmentations]:
# Apply transform
if hasattr(image, 'convert'):
# PIL Image
transformed = transform(image)
else:
# Already tensor
transformed = transform(image)
# Add batch dimension
transformed = transformed.unsqueeze(0).to(self.device)
# Get prediction
output = self.model(transformed)
predictions.append(F.softmax(output, dim=1))
# Average predictions
avg_prediction = torch.mean(torch.stack(predictions), dim=0)
return avg_prediction
def calculate_class_weights(train_targets, num_classes, method='balanced'):
"""
Calculate class weights for handling class imbalance.
Args:
train_targets: List of training targets
num_classes: Number of classes
method: 'balanced', 'inverse', or 'sqrt'
Returns:
class_weights: Tensor of class weights
"""
class_counts = np.bincount(train_targets, minlength=num_classes)
if method == 'balanced':
# sklearn's balanced method
total_samples = len(train_targets)
class_weights = total_samples / (num_classes * class_counts)
elif method == 'inverse':
# Simple inverse frequency
class_weights = 1.0 / class_counts
elif method == 'sqrt':
# Square root of inverse frequency
class_weights = 1.0 / np.sqrt(class_counts)
else:
raise ValueError(f"Unknown method: {method}")
# Normalize weights
class_weights = class_weights / class_weights.sum() * num_classes
return torch.tensor(class_weights, dtype=torch.float)
def get_advanced_scheduler(optimizer, method='cosine_warmup', total_epochs=50):
"""
Get advanced learning rate scheduler.
Args:
optimizer: PyTorch optimizer
method: Scheduler method
total_epochs: Total number of epochs
Returns:
scheduler: Learning rate scheduler
"""
if method == 'cosine_warmup':
from torch.optim.lr_scheduler import CosineAnnealingWarmRestarts
return CosineAnnealingWarmRestarts(optimizer, T_0=10, T_mult=2, eta_min=1e-7)
elif method == 'onecycle':
from torch.optim.lr_scheduler import OneCycleLR
return OneCycleLR(
optimizer,
max_lr=optimizer.param_groups[0]['lr'],
total_steps=total_epochs,
pct_start=0.3,
anneal_strategy='cos'
)
elif method == 'plateau':
from torch.optim.lr_scheduler import ReduceLROnPlateau
return ReduceLROnPlateau(
optimizer,
mode='max',
factor=0.5,
patience=3,
min_lr=1e-7,
verbose=True
)
else:
raise ValueError(f"Unknown scheduler method: {method}")
def apply_mixup_cutmix_probability():
"""
Randomly choose between Mixup and CutMix based on probability.
"""
return random.choice(['mixup', 'cutmix', 'none'])
|