File size: 2,818 Bytes
80e070e | 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 | """
Aggregate every band-combination run into one comparison table.
Reads, for each all_data/checkpoint*/ :
training_history.csv -> best val_dice_coef / val_iou_m (+ epoch)
run_config.json -> band list, n_bands (written by train.py)
and the matching all_data/result*/metrics.json -> test metrics (written by evaluate.py).
python results_table.py
Prints a table sorted by test_dice and writes all_data/runs_comparison.csv.
"""
import os
import glob
import json
import pandas as pd
ROOT = os.path.join(os.path.dirname(os.path.abspath(__file__)), "all_data")
def _best_val(hist_csv):
df = pd.read_csv(hist_csv)
if "val_dice_coef" not in df.columns or len(df) == 0:
return None
r = df.loc[df["val_dice_coef"].idxmax()]
return (round(float(r["val_dice_coef"]), 4),
round(float(r["val_iou_m"]), 4) if "val_iou_m" in df.columns else None,
int(r["epoch"]))
def main():
rows = []
for cdir in sorted(glob.glob(os.path.join(ROOT, "checkpoint*"))):
base = os.path.basename(cdir)
name = "all" if base == "checkpoint" else base[len("checkpoint_"):]
hist = os.path.join(cdir, "training_history.csv")
if not os.path.exists(hist):
continue
bv = _best_val(hist)
rc = os.path.join(cdir, "run_config.json")
cfg = json.load(open(rc)) if os.path.exists(rc) else {}
bands = cfg.get("bands")
n_bands = cfg.get("n_bands", len(bands) if bands else "?")
rdir = os.path.join(ROOT, "result" if name == "all" else f"result_{name}")
mp = os.path.join(rdir, "metrics.json")
m = json.load(open(mp)) if os.path.exists(mp) else {}
rows.append({
"run_name": name,
"n_bands": n_bands,
"val_dice": bv[0] if bv else None,
"val_iou": bv[1] if bv else None,
"best_ep": bv[2] if bv else None,
"test_dice": m.get("test_dice"),
"test_iou": m.get("test_iou"),
"precision": m.get("test_precision"),
"recall": m.get("test_recall"),
"img_mIoU": m.get("per_image_mean_iou"),
"bands": ", ".join(bands) if bands else "(all 35)",
})
if not rows:
print("No runs found under all_data/checkpoint*/ (train something first).")
return
df = pd.DataFrame(rows).sort_values(
"test_dice", ascending=False, na_position="last").reset_index(drop=True)
pd.set_option("display.width", 240)
pd.set_option("display.max_columns", 30)
print(df.drop(columns=["bands"]).to_string(index=False))
out = os.path.join(ROOT, "runs_comparison.csv")
df.to_csv(out, index=False)
print(f"\nFull table (with band lists) -> {out}")
if __name__ == "__main__":
main()
|