metadata/dataset_summary.json

{
    "dataset_name": "OCTID",
    "total_images": 572,
    "num_classes": 5,
    "classes": {
        "Normal": 206,
        "AMD": 55,
        "CSR": 102,
        "DR": 107,
        "MH": 102
    },
    "image_format": "JPEG",
    "image_size": "224x224",
    "image_mode": "RGB",
    "preprocessing": [
        "Aspect-ratio-preserving resize",
        "Center crop",
        "Non-Local Means denoising",
        "CLAHE"
    ],
    "normalization": "ImageNet normalization applied later in the PyTorch training pipeline",
    "splits": {
        "repetitions": 5,
        "train": 0.5,
        "validation": 0.2,
        "test": 0.3
    },
    "split_seeds": [
        42,
        52,
        62,
        72,
        82
    ]
}