Spaces:
Sleeping
Sleeping
File size: 9,149 Bytes
0b86da8 | 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 | import os
import sys
from pathlib import Path
# Add project root to sys.path
sys.path.append(str(Path(__file__).parent.parent))
import matplotlib # noqa: E402
matplotlib.use("Agg")
import matplotlib.pyplot as plt # noqa: E402
import mlflow # noqa: E402
import numpy as np # noqa: E402
import torch # noqa: E402
import torch.nn as nn # noqa: E402
import yaml # noqa: E402
from torch.utils.data import DataLoader # noqa: E402
from torchvision import transforms # noqa: E402
from tqdm import tqdm # noqa: E402
from src.dataset import TrashDataset # noqa: E402
from src.model import DeepCNN, ResNet18Transfer, SimpleCNN # noqa: E402
def load_config(config_path="config.yaml"):
with open(config_path, "r") as f:
return yaml.safe_load(f)
def get_device(config_device):
if config_device == "auto":
return "cuda" if torch.cuda.is_available() else "cpu"
return config_device
class Trainer:
"""
Handles the training and validation lifecycle of a model with MLflow tracking.
"""
def __init__(self, model, train_loader, val_loader, config, model_name):
self.config = config
self.model_name = model_name
self.device = get_device(config["device"])
self.model = model.to(self.device)
self.train_loader = train_loader
self.val_loader = val_loader
self.epochs = config["epochs"]
self.patience = config.get("patience", 5)
# Handle class imbalance with weights
y_train = np.load("data/processed/y_train.npy")
class_counts = np.bincount(y_train)
weights = 1.0 / class_counts
weights = torch.FloatTensor(weights).to(self.device)
self.criterion = nn.CrossEntropyLoss(weight=weights)
self.optimizer = torch.optim.Adam(model.parameters(), lr=config["lr"])
self.scheduler = torch.optim.lr_scheduler.CosineAnnealingLR(
self.optimizer, T_max=self.epochs
)
self.history = {"train_loss": [], "train_acc": [], "val_loss": [], "val_acc": []}
self.checkpoint_path = f"models/{model_name.lower()}_best.pth"
def train_epoch(self):
self.model.train()
running_loss = 0.0
correct = 0
total = 0
pbar = tqdm(self.train_loader, desc="Training", leave=False)
for images, labels in pbar:
images, labels = images.to(self.device), labels.to(self.device)
self.optimizer.zero_grad()
outputs = self.model(images)
loss = self.criterion(outputs, labels)
loss.backward()
self.optimizer.step()
running_loss += loss.item()
_, predicted = torch.max(outputs, 1)
total += labels.size(0)
correct += (predicted == labels).sum().item()
pbar.set_postfix({"loss": f"{loss.item():.4f}", "acc": f"{correct/total:.4f}"})
return running_loss / len(self.train_loader), correct / total
def validate(self):
self.model.eval()
running_loss = 0.0
correct = 0
total = 0
with torch.no_grad():
for images, labels in self.val_loader:
images, labels = images.to(self.device), labels.to(self.device)
outputs = self.model(images)
loss = self.criterion(outputs, labels)
running_loss += loss.item()
_, predicted = torch.max(outputs, 1)
total += labels.size(0)
correct += (predicted == labels).sum().item()
return running_loss / len(self.val_loader), correct / total
def plot_history(self):
save_path = f"models/plots/{self.model_name.lower()}_history.png"
os.makedirs(os.path.dirname(save_path), exist_ok=True)
epochs_range = range(1, len(self.history["train_loss"]) + 1)
plt.figure(figsize=(12, 5))
# Plot Loss
plt.subplot(1, 2, 1)
plt.plot(epochs_range, self.history["train_loss"], label="Train Loss")
plt.plot(epochs_range, self.history["val_loss"], label="Val Loss")
plt.title(f"{self.model_name} - Loss")
plt.xlabel("Epochs")
plt.ylabel("Loss")
plt.legend()
# Plot Accuracy
plt.subplot(1, 2, 2)
plt.plot(epochs_range, self.history["train_acc"], label="Train Acc")
plt.plot(epochs_range, self.history["val_acc"], label="Val Acc")
plt.title(f"{self.model_name} - Accuracy")
plt.xlabel("Epochs")
plt.ylabel("Accuracy")
plt.legend()
plt.tight_layout()
plt.savefig(save_path)
print(f"--> Training history plot saved to {save_path}")
mlflow.log_artifact(save_path)
def fit(self):
mlflow.set_experiment("Trash Classifier")
with mlflow.start_run(run_name=self.model_name):
mlflow.log_params(self.config)
mlflow.log_param("model_architecture", self.model_name)
print(f"\nStarting training for {self.model_name} on {self.device}...")
best_val_acc = 0.0
epochs_no_improve = 0
for epoch in range(self.epochs):
train_loss, train_acc = self.train_epoch()
val_loss, val_acc = self.validate()
self.scheduler.step()
self.history["train_loss"].append(train_loss)
self.history["train_acc"].append(train_acc)
self.history["val_loss"].append(val_loss)
self.history["val_acc"].append(val_acc)
mlflow.log_metric("train_loss", train_loss, step=epoch)
mlflow.log_metric("train_acc", train_acc, step=epoch)
mlflow.log_metric("val_loss", val_loss, step=epoch)
mlflow.log_metric("val_acc", val_acc, step=epoch)
mlflow.log_metric("lr", self.optimizer.param_groups[0]["lr"], step=epoch)
print(
f"Epoch [{epoch + 1}/{self.epochs}] "
f"Train Loss: {train_loss:.4f}, Acc: {train_acc:.4f} | "
f"Val Loss: {val_loss:.4f}, Acc: {val_acc:.4f} | "
f"LR: {self.optimizer.param_groups[0]['lr']:.6f}"
)
if val_acc > best_val_acc:
best_val_acc = val_acc
epochs_no_improve = 0
os.makedirs("models", exist_ok=True)
torch.save(self.model.state_dict(), self.checkpoint_path)
print(f"--> Saved best model for {self.model_name} with Val Acc: {val_acc:.4f}")
mlflow.log_artifact(self.checkpoint_path)
else:
epochs_no_improve += 1
if epochs_no_improve >= self.patience:
print(f"Early stopping triggered after {epoch + 1} epochs.")
break
self.plot_history()
return self.history
if __name__ == "__main__":
config = load_config()
train_transform = transforms.Compose(
[
transforms.ToPILImage(),
transforms.RandomResizedCrop(224, scale=(0.8, 1.0)),
transforms.RandomHorizontalFlip(),
transforms.RandomRotation(15),
transforms.ColorJitter(brightness=0.2, contrast=0.2, saturation=0.2),
transforms.ToTensor(),
transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
]
)
val_transform = transforms.Compose(
[
transforms.ToPILImage(),
transforms.ToTensor(),
transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
]
)
processed_dir = Path("data/processed")
if not (processed_dir / "X_train.npy").exists():
print("Error: Processed data not found. Please run src/dataset.py first.")
else:
train_ds = TrashDataset(
processed_dir / "X_train.npy", processed_dir / "y_train.npy", transform=train_transform
)
val_ds = TrashDataset(
processed_dir / "X_val.npy", processed_dir / "y_val.npy", transform=val_transform
)
train_loader = DataLoader(train_ds, batch_size=config["batch_size"], shuffle=True)
val_loader = DataLoader(val_ds, batch_size=config["batch_size"], shuffle=False)
num_classes = len(config["classes"])
models_to_train = {
"SimpleCNN": SimpleCNN(num_classes=num_classes),
"DeepCNN": DeepCNN(num_classes=num_classes),
"ResNet18": ResNet18Transfer(num_classes=num_classes, pretrained=True),
}
for name, model in models_to_train.items():
trainer = Trainer(model, train_loader, val_loader, config, name)
trainer.fit()
print("\nAll models trained. Starting comparison...")
from src.comparison import run_comparison
run_comparison()
|