XNet/datasets/PolypDetectionDataset/PolypDetectionDataset.py

from pathlib import Path
from typing import Callable, Optional, List, Dict, Any
from monai.data import Dataset


class PolypDetectionDataset(Dataset):
    """Polyp detection dataset for training and validation."""

    def __init__(
        self, root_dir: str, flag: str = "train", transform: Optional[Callable] = None
    ):
        """Initialize dataset.
        
        Args:
            root_dir: Root directory containing images, masks, and split files
            flag: Dataset split ('train' or 'val')
            transform: Optional transformations to apply
        """
        # Set up directory paths
        self.root_dir = Path(root_dir)
        self.images_dir = self.root_dir / "images"
        self.labels_dir = self.root_dir / "masks"
        self.flag = flag.lower()
        self.transform = transform

        # Validate split flag
        if self.flag not in ["train", "val"]:
            raise ValueError(f"flag must be 'train' or 'val', got '{self.flag}'")

        # Check if directories exist
        if not self.images_dir.exists():
            raise FileNotFoundError(f"Image directory does not exist: {self.images_dir}")
        if not self.labels_dir.exists():
            raise FileNotFoundError(f"Label directory does not exist: {self.labels_dir}")

        # Load image filenames from split file
        txt_file = self.root_dir / f"{self.flag}.txt"
        if not txt_file.exists():
            raise FileNotFoundError(
                f"{self.flag}.txt file not found in {self.root_dir}\n"
                f"Please ensure train.txt and val.txt files exist in this directory"
            )

        with open(txt_file, "r", encoding="utf-8") as f:
            self.images: List[str] = [
                line.strip() for line in f.readlines() if line.strip()
            ]

        # Labels have same filenames as images
        self.labels: List[str] = self.images.copy()

        # Create data list with image-label pairs
        data = [
            {"image": str(self.images_dir / img), "label": str(self.labels_dir / lbl)}
            for img, lbl in zip(self.images, self.labels)
        ]

        super().__init__(data=data, transform=transform)

    def __len__(self) -> int:
        """Return number of samples in dataset."""
        return len(self.images)

    def __getitem__(self, idx: int) -> Dict[str, Any]:
        """Get a single sample by index.
        
        Args:
            idx: Sample index
            
        Returns:
            Dictionary with 'image' and 'label' tensors
        """
        # Build file paths
        image_path = str(self.images_dir / self.images[idx])
        label_path = str(self.labels_dir / self.labels[idx])

        data = {"image": image_path, "label": label_path}

        # Apply transformations if specified
        if self.transform is not None:
            data = self.transform(data)

        return data

    def get_image_filename(self, idx: int) -> str:
        """Get image filename by index."""
        return self.images[idx]

    def get_label_filename(self, idx: int) -> str:
        """Get label filename by index."""
        return self.labels[idx]

    def get_dataset_info(self, dataset_name="CVC_300") -> Dict[str, Any]:
        """Get dataset information.
        
        Args:
            dataset_name: Name of the dataset
            
        Returns:
            Dictionary containing dataset metadata
        """
        return {
            "dataset_name": dataset_name,
            "split": self.flag,
            "num_samples": len(self),
            "root_dir": str(self.root_dir),
            "images_dir": str(self.images_dir),
            "labels_dir": str(self.labels_dir),
            "has_transform": self.transform is not None,
        }