{ "dataset_name": "Gatekeeper Binary Dataset (Preprocessed)", "purpose": "Binary gatekeeper for cervix detection before MedSigLip", "original_image_size": "448x448", "preprocessed_image_size": "224x224", "normalization": "[-1, 1] with mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5]", "total_samples": 11547, "classes": { "cervix": { "label": 1, "count": 5543 }, "not_cervix": { "label": 0, "count": 6004 } }, "split": { "train": "70%", "validation": "15%", "test": "15%" }, "stratified": true, "balance_ratio": "0.923", "model": "MobileNetV3-Small", "threshold": 0.70, "input_shape": [224, 224, 3], "note": "Images have been resized to 224x224 and normalized to [-1, 1]. This file describes the version used for training the gatekeeper model." }