Declare custom splits via README YAML (density-aware OOD splits)

README.md

@@ -8,6 +8,10 @@ configs:
     path: "data/configs/biome_India_Karnataka_train_DRY__ood_WET/id_test-*.parquet"
   - split: ood_test
     path: "data/configs/biome_India_Karnataka_train_DRY__ood_WET/ood_test-*.parquet"
+  - split: ood_same_density
+    path: "data/configs/biome_India_Karnataka_train_DRY__ood_WET/ood_same_density-*.parquet"
+  - split: ood_diff_density
+    path: "data/configs/biome_India_Karnataka_train_DRY__ood_WET/ood_diff_density-*.parquet"
 - config_name: biome_India_Karnataka_train_WET__ood_DRY
   data_files:
   - split: train
@@ -16,6 +20,10 @@ configs:
     path: "data/configs/biome_India_Karnataka_train_WET__ood_DRY/id_test-*.parquet"
   - split: ood_test
     path: "data/configs/biome_India_Karnataka_train_WET__ood_DRY/ood_test-*.parquet"
+  - split: ood_same_density
+    path: "data/configs/biome_India_Karnataka_train_WET__ood_DRY/ood_same_density-*.parquet"
+  - split: ood_diff_density
+    path: "data/configs/biome_India_Karnataka_train_WET__ood_DRY/ood_diff_density-*.parquet"
 - config_name: biome_India_Rajasthan_train_DRY__ood_WET
   data_files:
   - split: train
@@ -24,6 +32,10 @@ configs:
     path: "data/configs/biome_India_Rajasthan_train_DRY__ood_WET/id_test-*.parquet"
   - split: ood_test
     path: "data/configs/biome_India_Rajasthan_train_DRY__ood_WET/ood_test-*.parquet"
+  - split: ood_same_density
+    path: "data/configs/biome_India_Rajasthan_train_DRY__ood_WET/ood_same_density-*.parquet"
+  - split: ood_diff_density
+    path: "data/configs/biome_India_Rajasthan_train_DRY__ood_WET/ood_diff_density-*.parquet"
 - config_name: biome_India_Rajasthan_train_WET__ood_DRY
   data_files:
   - split: train
@@ -32,6 +44,10 @@ configs:
     path: "data/configs/biome_India_Rajasthan_train_WET__ood_DRY/id_test-*.parquet"
   - split: ood_test
     path: "data/configs/biome_India_Rajasthan_train_WET__ood_DRY/ood_test-*.parquet"
+  - split: ood_same_density
+    path: "data/configs/biome_India_Rajasthan_train_WET__ood_DRY/ood_same_density-*.parquet"
+  - split: ood_diff_density
+    path: "data/configs/biome_India_Rajasthan_train_WET__ood_DRY/ood_diff_density-*.parquet"
 - config_name: biome_US_California_train_DRY__ood_WET
   data_files:
   - split: train
@@ -40,6 +56,10 @@ configs:
     path: "data/configs/biome_US_California_train_DRY__ood_WET/id_test-*.parquet"
   - split: ood_test
     path: "data/configs/biome_US_California_train_DRY__ood_WET/ood_test-*.parquet"
+  - split: ood_same_density
+    path: "data/configs/biome_US_California_train_DRY__ood_WET/ood_same_density-*.parquet"
+  - split: ood_diff_density
+    path: "data/configs/biome_US_California_train_DRY__ood_WET/ood_diff_density-*.parquet"
 - config_name: biome_US_California_train_WET__ood_DRY
   data_files:
   - split: train
@@ -48,6 +68,10 @@ configs:
     path: "data/configs/biome_US_California_train_WET__ood_DRY/id_test-*.parquet"
   - split: ood_test
     path: "data/configs/biome_US_California_train_WET__ood_DRY/ood_test-*.parquet"
+  - split: ood_same_density
+    path: "data/configs/biome_US_California_train_WET__ood_DRY/ood_same_density-*.parquet"
+  - split: ood_diff_density
+    path: "data/configs/biome_US_California_train_WET__ood_DRY/ood_diff_density-*.parquet"
 - config_name: in_state_train_Karnataka__ood_Rajasthan
   data_files:
   - split: train
@@ -56,6 +80,10 @@ configs:
     path: "data/configs/in_state_train_Karnataka__ood_Rajasthan/id_test-*.parquet"
   - split: ood_test
     path: "data/configs/in_state_train_Karnataka__ood_Rajasthan/ood_test-*.parquet"
+  - split: ood_same_density
+    path: "data/configs/in_state_train_Karnataka__ood_Rajasthan/ood_same_density-*.parquet"
+  - split: ood_diff_density
+    path: "data/configs/in_state_train_Karnataka__ood_Rajasthan/ood_diff_density-*.parquet"
 - config_name: in_state_train_Rajasthan__ood_Karnataka
   data_files:
   - split: train
@@ -64,6 +92,10 @@ configs:
     path: "data/configs/in_state_train_Rajasthan__ood_Karnataka/id_test-*.parquet"
   - split: ood_test
     path: "data/configs/in_state_train_Rajasthan__ood_Karnataka/ood_test-*.parquet"
+  - split: ood_same_density
+    path: "data/configs/in_state_train_Rajasthan__ood_Karnataka/ood_same_density-*.parquet"
+  - split: ood_diff_density
+    path: "data/configs/in_state_train_Rajasthan__ood_Karnataka/ood_diff_density-*.parquet"
 - config_name: intl_train_IN__ood_US
   data_files:
   - split: train
@@ -72,6 +104,10 @@ configs:
     path: "data/configs/intl_train_IN__ood_US/id_test-*.parquet"
   - split: ood_test
     path: "data/configs/intl_train_IN__ood_US/ood_test-*.parquet"
+  - split: ood_same_density
+    path: "data/configs/intl_train_IN__ood_US/ood_same_density-*.parquet"
+  - split: ood_diff_density
+    path: "data/configs/intl_train_IN__ood_US/ood_diff_density-*.parquet"
 - config_name: intl_train_US__ood_IN
   data_files:
   - split: train
@@ -80,6 +116,10 @@ configs:
     path: "data/configs/intl_train_US__ood_IN/id_test-*.parquet"
   - split: ood_test
     path: "data/configs/intl_train_US__ood_IN/ood_test-*.parquet"
+  - split: ood_same_density
+    path: "data/configs/intl_train_US__ood_IN/ood_same_density-*.parquet"
+  - split: ood_diff_density
+    path: "data/configs/intl_train_US__ood_IN/ood_diff_density-*.parquet"
 ---
 
 # Tree Distribution Shift — Satellite Tree Detection (COCO + HF Datasets)

tools/update_readme_yaml.py

@@ -0,0 +1,50 @@
+from __future__ import annotations
+from pathlib import Path
+import re
+
+REPO_ROOT = Path(__file__).resolve().parents[1]
+README = REPO_ROOT / "README.md"
+CONFIGS_DIR = REPO_ROOT / "data" / "configs"
+
+SPLITS = [
+    "train",
+    "id_test",
+    "ood_test",
+    "ood_same_density",
+    "ood_diff_density",
+]
+
+def has_files(cfg_dir: Path, split: str) -> bool:
+    return any(cfg_dir.glob(f"{split}-*.parquet"))
+
+def build_yaml() -> str:
+    cfg_dirs = sorted([p for p in CONFIGS_DIR.iterdir() if p.is_dir()], key=lambda p: p.name)
+    lines = ["---", "configs:"]
+    for cfg in cfg_dirs:
+        present = [s for s in SPLITS if has_files(cfg, s)]
+        # always include at least train/id_test/ood_test, but only write what's present on disk
+        lines.append(f"- config_name: {cfg.name}")
+        lines.append("  data_files:")
+        for s in present:
+            lines.append(f"  - split: {s}")
+            lines.append(f"    path: \"data/configs/{cfg.name}/{s}-*.parquet\"")
+    lines.append("---")
+    return "\n".join(lines) + "\n"
+
+def strip_existing_yaml(text: str) -> str:
+    # Remove existing top YAML block if present
+    if text.startswith("---"):
+        m = re.search(r"^---\s*\n.*?\n---\s*\n", text, flags=re.S)
+        if m:
+            return text[m.end():]
+    return text
+
+def main():
+    old = README.read_text() if README.exists() else ""
+    body = strip_existing_yaml(old).lstrip("\n")
+    new = build_yaml() + "\n" + body
+    README.write_text(new)
+    print(f"Wrote YAML configs to {README}")
+
+if __name__ == "__main__":
+    main()