File size: 9,859 Bytes
ab93c59 5265d5f ab93c59 5265d5f ab93c59 5265d5f ab93c59 5265d5f ab93c59 5265d5f ab93c59 5265d5f ab93c59 5265d5f ab93c59 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 | #!/usr/bin/env python3
"""Create a MILK10k-style dataset folder with original + synthetic paired augmentations."""
from __future__ import annotations
import argparse
import csv
import shutil
from pathlib import Path
LABEL_COLUMNS = [
"AKIEC",
"BCC",
"BEN_OTH",
"BKL",
"DF",
"INF",
"MAL_OTH",
"MEL",
"NV",
"SCCKA",
"VASC",
]
NEUTRAL_METADATA = {
"age_approx": "",
"sex": "unknown",
"skin_tone_class": "",
"site": "unknown",
}
def parse_args() -> argparse.Namespace:
parser = argparse.ArgumentParser(description="Materialize original + synthetic data as a MILK10k-style dataset.")
parser.add_argument("--input-dir", type=Path, required=True, help="Original MILK10k_Training_Input folder.")
parser.add_argument("--metadata-csv", type=Path, required=True, help="Original MILK10k_Training_Metadata.csv.")
parser.add_argument("--groundtruth-csv", type=Path, required=True, help="Original MILK10k_Training_GroundTruth.csv.")
parser.add_argument("--augmentation-manifest", type=Path, required=True, help="Paired augmentation manifest to add.")
parser.add_argument("--output-dir", type=Path, required=True, help="Output MILK10k-style dataset folder.")
parser.add_argument(
"--symlink",
action="store_true",
help="Symlink images instead of copying. Default is copy, which is easier to move/use later.",
)
parser.add_argument(
"--synthetic-metadata",
choices=["source", "neutral"],
default="source",
help="Metadata for synthetic rows. source copies source lesion metadata; neutral writes unknown/0 values.",
)
parser.add_argument("--overwrite", action="store_true", help="Overwrite output CSV files and existing image links.")
return parser.parse_args()
def read_rows(path: Path) -> list[dict[str, str]]:
with path.open(newline="") as f:
return list(csv.DictReader(f))
def write_rows(path: Path, rows: list[dict[str, str]], fieldnames: list[str]) -> None:
path.parent.mkdir(parents=True, exist_ok=True)
with path.open("w", newline="") as f:
writer = csv.DictWriter(f, fieldnames=fieldnames)
writer.writeheader()
writer.writerows(rows)
def link_or_copy(src: Path, dst: Path, copy_file: bool, overwrite: bool) -> None:
if not src.exists():
raise FileNotFoundError(f"Source image not found: {src}")
dst.parent.mkdir(parents=True, exist_ok=True)
if dst.exists() or dst.is_symlink():
if not overwrite:
return
dst.unlink()
if copy_file:
shutil.copy2(src, dst)
else:
dst.symlink_to(src.resolve())
def original_image_path(input_dir: Path, row: dict[str, str]) -> Path:
return input_dir / row["lesion_id"] / f"{row['isic_id']}.jpg"
def synthetic_metadata_row(
columns: list[str],
lesion_id: str,
image_type: str,
isic_id: str,
) -> dict[str, str]:
row = {column: "" for column in columns}
row["lesion_id"] = lesion_id
row["image_type"] = image_type
row["isic_id"] = isic_id
if "attribution" in row:
row["attribution"] = "Stable Diffusion synthetic augmentation"
if "copyright_license" in row:
row["copyright_license"] = "synthetic"
if "image_manipulation" in row:
row["image_manipulation"] = "synthetic"
for key, value in NEUTRAL_METADATA.items():
if key in row:
row[key] = value
for column in columns:
if column.startswith("MONET_"):
row[column] = "0"
return row
def source_metadata_row(
source_row: dict[str, str] | None,
columns: list[str],
lesion_id: str,
image_type: str,
isic_id: str,
) -> dict[str, str]:
if source_row is None:
return synthetic_metadata_row(columns, lesion_id, image_type, isic_id)
row = {column: source_row.get(column, "") for column in columns}
row["lesion_id"] = lesion_id
row["image_type"] = image_type
row["isic_id"] = isic_id
if "attribution" in row:
row["attribution"] = "Stable Diffusion synthetic augmentation"
if "copyright_license" in row:
row["copyright_license"] = "synthetic"
if "image_manipulation" in row:
row["image_manipulation"] = "synthetic_from_source"
return row
def synthetic_groundtruth_row(lesion_id: str, class_name: str, columns: list[str]) -> dict[str, str]:
row = {column: "0.0" for column in columns}
row["lesion_id"] = lesion_id
for class_column in LABEL_COLUMNS:
if class_column in row:
row[class_column] = "1.0" if class_column == class_name else "0.0"
return row
def materialize_original_images(input_dir: Path, output_input_dir: Path, metadata_rows: list[dict[str, str]], copy_file: bool, overwrite: bool) -> None:
for row in metadata_rows:
src = original_image_path(input_dir, row)
dst = output_input_dir / row["lesion_id"] / f"{row['isic_id']}.jpg"
link_or_copy(src, dst, copy_file, overwrite)
def materialize_synthetic_rows(
manifest_rows: list[dict[str, str]],
metadata_columns: list[str],
groundtruth_columns: list[str],
source_metadata_by_key: dict[tuple[str, str], dict[str, str]],
synthetic_metadata_mode: str,
output_input_dir: Path,
copy_file: bool,
overwrite: bool,
) -> tuple[list[dict[str, str]], list[dict[str, str]]]:
metadata_rows = []
groundtruth_rows = []
seen_lesions = set()
for row in manifest_rows:
lesion_id = row["synthetic_lesion_id"]
clinical_isic_id = row.get("clinical_synthetic_isic_id") or f"{lesion_id}__clinical"
dermoscopic_isic_id = row.get("dermoscopic_synthetic_isic_id") or f"{lesion_id}__dermoscopic"
clinical_dst = output_input_dir / lesion_id / f"{clinical_isic_id}.jpg"
dermoscopic_dst = output_input_dir / lesion_id / f"{dermoscopic_isic_id}.jpg"
link_or_copy(Path(row["clinical_generated_path"]), clinical_dst, copy_file, overwrite)
link_or_copy(Path(row["dermoscopic_generated_path"]), dermoscopic_dst, copy_file, overwrite)
if synthetic_metadata_mode == "source":
source_lesion_id = row.get("source_lesion_id", "")
clinical_source_isic_id = row.get("clinical_source_isic_id", "")
dermoscopic_source_isic_id = row.get("dermoscopic_source_isic_id", "")
metadata_rows.append(
source_metadata_row(
source_metadata_by_key.get((source_lesion_id, clinical_source_isic_id)),
metadata_columns,
lesion_id,
"clinical: close-up",
clinical_isic_id,
)
)
metadata_rows.append(
source_metadata_row(
source_metadata_by_key.get((source_lesion_id, dermoscopic_source_isic_id)),
metadata_columns,
lesion_id,
"dermoscopic",
dermoscopic_isic_id,
)
)
else:
metadata_rows.append(synthetic_metadata_row(metadata_columns, lesion_id, "clinical: close-up", clinical_isic_id))
metadata_rows.append(synthetic_metadata_row(metadata_columns, lesion_id, "dermoscopic", dermoscopic_isic_id))
if lesion_id not in seen_lesions:
groundtruth_rows.append(synthetic_groundtruth_row(lesion_id, row["class_name"], groundtruth_columns))
seen_lesions.add(lesion_id)
return metadata_rows, groundtruth_rows
def main() -> None:
args = parse_args()
input_dir = args.input_dir.expanduser().resolve()
metadata_csv = args.metadata_csv.expanduser().resolve()
groundtruth_csv = args.groundtruth_csv.expanduser().resolve()
augmentation_manifest = args.augmentation_manifest.expanduser().resolve()
output_dir = args.output_dir.expanduser().resolve()
output_input_dir = output_dir / "MILK10k_Training_Input"
metadata_rows = read_rows(metadata_csv)
groundtruth_rows = read_rows(groundtruth_csv)
manifest_rows = read_rows(augmentation_manifest)
if not manifest_rows:
raise ValueError(f"No augmentation rows found: {augmentation_manifest}")
metadata_columns = list(metadata_rows[0].keys())
groundtruth_columns = list(groundtruth_rows[0].keys())
source_metadata_by_key = {(row["lesion_id"], row["isic_id"]): row for row in metadata_rows}
copy_file = not args.symlink
materialize_original_images(input_dir, output_input_dir, metadata_rows, copy_file, args.overwrite)
synthetic_metadata_rows, synthetic_groundtruth_rows = materialize_synthetic_rows(
manifest_rows,
metadata_columns,
groundtruth_columns,
source_metadata_by_key,
args.synthetic_metadata,
output_input_dir,
copy_file,
args.overwrite,
)
all_metadata_rows = metadata_rows + synthetic_metadata_rows
all_groundtruth_rows = groundtruth_rows + synthetic_groundtruth_rows
write_rows(output_dir / "MILK10k_Training_Metadata.csv", all_metadata_rows, metadata_columns)
write_rows(output_dir / "MILK10k_Training_GroundTruth.csv", all_groundtruth_rows, groundtruth_columns)
print("Materialized augmented MILK10k dataset")
print(f" output_dir: {output_dir}")
print(f" original metadata rows: {len(metadata_rows)}")
print(f" synthetic metadata rows: {len(synthetic_metadata_rows)}")
print(f" original groundtruth rows: {len(groundtruth_rows)}")
print(f" synthetic groundtruth rows: {len(synthetic_groundtruth_rows)}")
print(f" image mode: {'symlink' if args.symlink else 'copy'}")
print(f" synthetic metadata: {args.synthetic_metadata}")
print("")
print("Use this for training:")
print(f" --data-dir {output_dir}")
if __name__ == "__main__":
main()
|