feat: add ColorJitter augmentation to training data

Co-authored-by: aider (gemini/gemini-2.5-pro-preview-05-06) <aider@aider.chat>

train.py

@@ -1,7 +1,7 @@
 import torch
 import torch.nn as nn
 import torch.optim as optim
-from torch.utils.data import DataLoader, random_split, WeightedRandomSampler
+from torch.utils.data import DataLoader, random_split, WeightedRandomSampler, Dataset
 from torchvision import datasets, transforms
 from PIL import Image, ImageDraw
 import os
@@ -40,6 +40,26 @@ def check_crop():
     print(f"Debug image saved to '{output_path}'.")
 
 
+class TransformedSubset(Dataset):
+    """
+    A wrapper for a Subset that allows applying a transform.
+    This is necessary because a transform cannot be applied to a Subset directly.
+    """
+    def __init__(self, subset, transform=None):
+        self.subset = subset
+        self.transform = transform
+
+    def __getitem__(self, index):
+        # The subset returns the data from the original dataset (img, label)
+        img, label = self.subset[index]
+        if self.transform:
+            img = self.transform(img)
+        return img, label
+
+    def __len__(self):
+        return len(self.subset)
+
+
 def train_model():
     # --- Hyperparameters and Configuration ---
     DATA_DIR = 'data/labelled'
@@ -49,25 +69,37 @@ def train_model():
     LEARNING_RATE = 0.001
 
     # --- Data Preparation ---
-    # Define transforms
+    # Define separate transforms for training (with augmentation) and validation (without)
-    data_transforms = transforms.Compose([
+    train_transforms = transforms.Compose([
+        CropLowerRightTriangle(triangle_width=TRIANGLE_CROP_WIDTH, triangle_height=TRIANGLE_CROP_HEIGHT),
+        transforms.ColorJitter(brightness=0.3, contrast=0.3, saturation=0.2),
+        transforms.Resize((RESIZE_DIM, RESIZE_DIM)),
+        transforms.ToTensor(),
+        transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
+    ])
+
+    val_transforms = transforms.Compose([
         CropLowerRightTriangle(triangle_width=TRIANGLE_CROP_WIDTH, triangle_height=TRIANGLE_CROP_HEIGHT),
         transforms.Resize((RESIZE_DIM, RESIZE_DIM)),
         transforms.ToTensor(),
         transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
     ])
 
-    # Load dataset with ImageFolder
+    # Load dataset without transforms, as they will be applied to subsets
-    full_dataset = datasets.ImageFolder(DATA_DIR, transform=data_transforms)
+    untransformed_dataset = datasets.ImageFolder(DATA_DIR)
     
     # Split into training and validation sets
-    train_size = int(0.8 * len(full_dataset))
+    train_size = int(0.8 * len(untransformed_dataset))
-    val_size = len(full_dataset) - train_size
+    val_size = len(untransformed_dataset) - train_size
-    train_dataset, val_dataset = random_split(full_dataset, [train_size, val_size])
+    train_subset, val_subset = random_split(untransformed_dataset, [train_size, val_size])
+
+    # Apply the respective transforms to the subsets using our wrapper
+    train_dataset = TransformedSubset(train_subset, transform=train_transforms)
+    val_dataset = TransformedSubset(val_subset, transform=val_transforms)
 
     # --- Handle Class Imbalance ---
-    # Get labels for training set
+    # Get labels for training set from the subset indices
-    train_labels = [full_dataset.targets[i] for i in train_dataset.indices]
+    train_labels = [untransformed_dataset.targets[i] for i in train_subset.indices]
     
     # Get class counts
     class_counts = torch.bincount(torch.tensor(train_labels))