Spaces:
Sleeping
Sleeping
File size: 9,276 Bytes
40348b6 abd02d5 |
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 |
"""
Bird classification model architectures with overfitting prevention.
"""
import torch
import torch.nn as nn
import torch.nn.functional as F
from torchvision import models
from typing import Optional
# Try to import EfficientNet
try:
from efficientnet_pytorch import EfficientNet
EFFICIENTNET_AVAILABLE = True
except ImportError:
EFFICIENTNET_AVAILABLE = False
print("EfficientNet not available. Install with: pip install efficientnet-pytorch")
class BirdClassifier(nn.Module):
"""
Bird classification model with ResNet backbone and overfitting prevention.
"""
def __init__(self, num_classes: int, architecture: str = 'resnet50',
pretrained: bool = True, dropout_rate: float = 0.5,
freeze_backbone: bool = False):
"""
Initialize the bird classifier.
Args:
num_classes: Number of bird classes
architecture: Backbone architecture ('resnet50', 'resnet18', 'efficientnet_b0')
pretrained: Whether to use pretrained weights
dropout_rate: Dropout rate for regularization
freeze_backbone: Whether to freeze backbone weights
"""
super(BirdClassifier, self).__init__()
self.num_classes = num_classes
self.dropout_rate = dropout_rate
# Choose backbone architecture
if architecture == 'resnet50':
self.backbone = models.resnet50(pretrained=pretrained)
num_features = self.backbone.fc.in_features
self.backbone.fc = nn.Identity() # Remove original classifier
elif architecture == 'resnet18':
self.backbone = models.resnet18(pretrained=pretrained)
num_features = self.backbone.fc.in_features
self.backbone.fc = nn.Identity()
elif architecture == 'resnet101':
self.backbone = models.resnet101(pretrained=pretrained)
num_features = self.backbone.fc.in_features
self.backbone.fc = nn.Identity()
elif architecture == 'efficientnet_b0':
self.backbone = models.efficientnet_b0(pretrained=pretrained)
num_features = self.backbone.classifier[1].in_features
self.backbone.classifier = nn.Identity()
elif architecture in ['efficientnet_b1', 'efficientnet_b2', 'efficientnet_b3', 'efficientnet_b4'] and EFFICIENTNET_AVAILABLE:
model_name = architecture.replace('_', '-')
if pretrained:
self.backbone = EfficientNet.from_pretrained(model_name)
else:
self.backbone = EfficientNet.from_name(model_name)
num_features = self.backbone._fc.in_features
self.backbone._fc = nn.Identity()
else:
raise ValueError(f"Unsupported architecture: {architecture}")
# Freeze backbone if requested
if freeze_backbone:
for param in self.backbone.parameters():
param.requires_grad = False
# Enhanced classifier head with batch normalization and progressive dimension reduction
# Optimized regularization for Stage 2 performance (76.74% accuracy)
self.classifier = nn.Sequential(
nn.Dropout(p=dropout_rate * 0.6), # Stage 2 optimization: 0.3 * 0.6 = 0.18
nn.Linear(num_features, 512), # Optimized size
nn.BatchNorm1d(512),
nn.ReLU(inplace=True),
nn.Dropout(p=dropout_rate * 0.5), # Stage 2 optimization: 0.3 * 0.5 = 0.15
nn.Linear(512, 256),
nn.BatchNorm1d(256),
nn.ReLU(inplace=True),
nn.Dropout(p=dropout_rate * 0.3), # Stage 2 optimization: 0.3 * 0.3 = 0.09
nn.Linear(256, num_classes)
)
# Initialize weights
self._initialize_weights()
def _initialize_weights(self):
"""Initialize classifier weights with better initialization."""
for m in self.classifier.modules():
if isinstance(m, nn.Linear):
nn.init.xavier_uniform_(m.weight, gain=nn.init.calculate_gain('relu'))
if m.bias is not None:
nn.init.constant_(m.bias, 0)
elif isinstance(m, nn.BatchNorm1d):
nn.init.constant_(m.weight, 1)
nn.init.constant_(m.bias, 0)
def forward(self, x):
"""Forward pass."""
features = self.backbone(x)
output = self.classifier(features)
return output
class LightweightBirdClassifier(nn.Module):
"""
Lightweight CNN model for bird classification with batch normalization.
"""
def __init__(self, num_classes: int, dropout_rate: float = 0.5):
"""
Initialize lightweight classifier.
Args:
num_classes: Number of bird classes
dropout_rate: Dropout rate for regularization
"""
super(LightweightBirdClassifier, self).__init__()
self.features = nn.Sequential(
# Block 1
nn.Conv2d(3, 32, kernel_size=3, padding=1),
nn.BatchNorm2d(32),
nn.ReLU(inplace=True),
nn.Conv2d(32, 32, kernel_size=3, padding=1),
nn.BatchNorm2d(32),
nn.ReLU(inplace=True),
nn.MaxPool2d(2, 2),
nn.Dropout2d(p=dropout_rate/2),
# Block 2
nn.Conv2d(32, 64, kernel_size=3, padding=1),
nn.BatchNorm2d(64),
nn.ReLU(inplace=True),
nn.Conv2d(64, 64, kernel_size=3, padding=1),
nn.BatchNorm2d(64),
nn.ReLU(inplace=True),
nn.MaxPool2d(2, 2),
nn.Dropout2d(p=dropout_rate/2),
# Block 3
nn.Conv2d(64, 128, kernel_size=3, padding=1),
nn.BatchNorm2d(128),
nn.ReLU(inplace=True),
nn.Conv2d(128, 128, kernel_size=3, padding=1),
nn.BatchNorm2d(128),
nn.ReLU(inplace=True),
nn.MaxPool2d(2, 2),
nn.Dropout2d(p=dropout_rate/2),
# Block 4
nn.Conv2d(128, 256, kernel_size=3, padding=1),
nn.BatchNorm2d(256),
nn.ReLU(inplace=True),
nn.Conv2d(256, 256, kernel_size=3, padding=1),
nn.BatchNorm2d(256),
nn.ReLU(inplace=True),
nn.AdaptiveAvgPool2d((1, 1)),
)
self.classifier = nn.Sequential(
nn.Flatten(),
nn.Dropout(p=dropout_rate),
nn.Linear(256, 128),
nn.ReLU(inplace=True),
nn.Dropout(p=dropout_rate),
nn.Linear(128, num_classes)
)
self._initialize_weights()
def _initialize_weights(self):
"""Initialize model weights."""
for m in self.modules():
if isinstance(m, nn.Conv2d):
nn.init.kaiming_normal_(m.weight, mode='fan_out', nonlinearity='relu')
if m.bias is not None:
nn.init.constant_(m.bias, 0)
elif isinstance(m, nn.BatchNorm2d):
nn.init.constant_(m.weight, 1)
nn.init.constant_(m.bias, 0)
elif isinstance(m, nn.Linear):
nn.init.xavier_uniform_(m.weight)
nn.init.constant_(m.bias, 0)
def forward(self, x):
"""Forward pass."""
x = self.features(x)
x = self.classifier(x)
return x
def create_model(num_classes: int, model_type: str = 'resnet50',
pretrained: bool = True, dropout_rate: float = 0.5,
freeze_backbone: bool = False) -> nn.Module:
"""
Create a bird classification model.
Args:
num_classes: Number of bird classes
model_type: Type of model ('resnet50', 'resnet18', 'efficientnet_b0', 'lightweight')
pretrained: Whether to use pretrained weights
dropout_rate: Dropout rate for regularization
freeze_backbone: Whether to freeze backbone weights (ignored for lightweight model)
Returns:
PyTorch model
"""
if model_type == 'lightweight':
return LightweightBirdClassifier(num_classes, dropout_rate)
else:
return BirdClassifier(num_classes, model_type, pretrained,
dropout_rate, freeze_backbone)
class ModelEnsemble(nn.Module):
"""
Ensemble of multiple models for improved performance.
"""
def __init__(self, models_list: list):
"""
Initialize model ensemble.
Args:
models_list: List of trained models to ensemble
"""
super(ModelEnsemble, self).__init__()
self.models = nn.ModuleList(models_list)
def forward(self, x):
"""Forward pass through all models and average predictions."""
predictions = []
for model in self.models:
with torch.no_grad():
pred = F.softmax(model(x), dim=1)
predictions.append(pred)
# Average predictions
ensemble_pred = torch.stack(predictions, dim=0).mean(dim=0)
return torch.log(ensemble_pred + 1e-8) # Convert back to log probabilities |