Add Kvasir-SEG dataset

scripts/datasets/check_kvasir.py

@@ -0,0 +1,21 @@
+from torch_em.util.debug import check_loader
+from torch_em.data.datasets.medical import get_kvasir_loader
+
+
+ROOT = "/media/anwai/ANWAI/data/kvasir"
+
+
+def check_kvasir():
+    loader = get_kvasir_loader(
+        path=ROOT,
+        patch_shape=(512, 512),
+        batch_size=2,
+        download=True,
+        resize_inputs=False,
+    )
+
+    check_loader(loader, 8)
+
+
+if __name__ == "__main__":
+    check_kvasir()

torch_em/data/datasets/medical/__init__.py

@@ -2,5 +2,6 @@
 from .btcv import get_btcv_dataset, get_btcv_loader
 from .camus import get_camus_dataset, get_camus_loader
 from .drive import get_drive_dataset, get_drive_loader
+from .kvasir import get_kvasir_dataset, get_kvasir_loader
 from .papila import get_papila_dataset, get_papila_loader
 from .siim_acr import get_siim_acr_dataset, get_siim_acr_loader

torch_em/data/datasets/medical/kvasir.py

@@ -0,0 +1,110 @@
+import os
+from glob import glob
+from tqdm import tqdm
+from pathlib import Path
+from typing import Union, Tuple
+
+import numpy as np
+import imageio.v3 as imageio
+
+import torch_em
+from torch_em.transform.generic import ResizeInputs
+
+from .. import util
+from ... import ImageCollectionDataset
+
+
+URL = "https://datasets.simula.no/downloads/kvasir-seg.zip"
+CHECKSUM = "03b30e21d584e04facf49397a2576738fd626815771afbbf788f74a7153478f7"
+
+
+def get_kvasir_data(path, download):
+    os.makedirs(path, exist_ok=True)
+
+    data_dir = os.path.join(path, "Kvasir-SEG")
+    if os.path.exists(data_dir):
+        return data_dir
+
+    zip_path = os.path.join(path, "kvasir-seg.zip")
+    util.download_source(path=zip_path, url=URL, download=download, checksum=CHECKSUM)
+    util.unzip(zip_path=zip_path, dst=path)
+
+    return data_dir
+
+
+def _get_kvasir_paths(path, download):
+    data_dir = get_kvasir_data(path=path, download=download)
+
+    image_paths = sorted(glob(os.path.join(data_dir, "images", "*.jpg")))
+    gt_paths = sorted(glob(os.path.join(data_dir, "masks", "*.jpg")))
+
+    neu_gt_dir = os.path.join(data_dir, "masks", "preprocessed")
+    os.makedirs(neu_gt_dir, exist_ok=True)
+
+    neu_gt_paths = []
+    for gt_path in tqdm(gt_paths):
+        neu_gt_path = os.path.join(neu_gt_dir, f"{Path(gt_path).stem}.tif")
+        neu_gt_paths.append(neu_gt_path)
+        if os.path.exists(neu_gt_path):
+            continue
+
+        gt = imageio.imread(gt_path)
+        gt = np.mean(gt, axis=-1)
+        gt = (gt >= 240).astype("uint8")
+        imageio.imwrite(neu_gt_path, gt)
+
+    return image_paths, neu_gt_paths
+
+
+def get_kvasir_dataset(
+    path: Union[os.PathLike, str],
+    patch_shape: Tuple[int, int],
+    resize_inputs: bool = False,
+    download: bool = False,
+    **kwargs
+):
+    """Dataset for polyp segmentation in colonoscopy images.
+
+    The dataset is located at https://datasets.simula.no/kvasir-seg/
+
+    This dataset is from Jha et al. - https://doi.org/10.1007/978-3-030-37734-2_37
+    Please cite it if you use this dataset for a publication.
+    """
+    image_paths, gt_paths = _get_kvasir_paths(path=path, download=download)
+
+    if resize_inputs:
+        raw_trafo = ResizeInputs(target_shape=patch_shape, is_rgb=True)
+        label_trafo = ResizeInputs(target_shape=patch_shape, is_label=True)
+        patch_shape = None
+    else:
+        patch_shape = patch_shape
+        raw_trafo, label_trafo = None, None
+
+    dataset = ImageCollectionDataset(
+        raw_image_paths=image_paths,
+        label_image_paths=gt_paths,
+        patch_shape=patch_shape,
+        raw_transform=raw_trafo,
+        label_transform=label_trafo,
+        **kwargs
+    )
+
+    return dataset
+
+
+def get_kvasir_loader(
+    path: Union[os.PathLike, str],
+    patch_shape: Tuple[int, int],
+    batch_size: int,
+    resize_inputs: bool = False,
+    download: bool = False,
+    **kwargs
+):
+    """Dataloader for polyp segmentation in colonoscopy images. See `get_kvasir_dataset` for details.
+    """
+    ds_kwargs, loader_kwargs = util.split_kwargs(torch_em.default_segmentation_dataset, **kwargs)
+    dataset = get_kvasir_dataset(
+        path=path, patch_shape=patch_shape, resize_inputs=resize_inputs, download=download, **ds_kwargs
+    )
+    loader = torch_em.get_data_loader(dataset=dataset, batch_size=batch_size, **loader_kwargs)
+    return loader