example.py

from ODDataset import ObjectDetectionDataset
from torch.utils.data import DataLoader
from torchvision import transforms



default_transform = transforms.Compose([
    transforms.ToTensor(),
])

def collate_fn(batch):
    return tuple(zip(*batch))

# Expected structure:
# mydataset/
# ├── dimensions/
# │   ├── image_001.jpg
# │   └── image_002.jpg
# └── labels.json
dataset = ObjectDetectionDataset(
    image_folder="mydataset/dimensions",
    labels_json_path="mydataset/labels.json",
    transform=default_transform,
)

loader = DataLoader(
    dataset,
    batch_size=4,
    shuffle=True,
    collate_fn=collate_fn,
)

for images, targets in loader:
    print(f"Images batch shape: {images[0].shape}")
    print(f"Targets: {targets, }")
    break  # just the first batch for testing