AIChamcong / src /datasets /generate_split_files.py
quangdat05's picture
Clean deploy with datetime parsing fixes
1cd0545
Raw History Blame Contribute Delete
1.95 kB
import os
AFDB_FACE_ROOT = "dataset/RWFRD/AFDB_face_dataset/AFDB_face_dataset"
AFDB_MASKED_ROOT = "dataset/RWFRD/AFDB_masked_face_dataset/AFDB_masked_face_dataset"
OUTPUT_DIR = "dataset/RWFRD"
SPLIT_RATIO = 0.8
def main():
print("=" * 70)
print("GENERATING TRAIN/TEST IDENTITY SPLIT FILES")
print("=" * 70)
# 1. Scan all unique identity folder names across both roots
identity_names = set()
for root in [AFDB_FACE_ROOT, AFDB_MASKED_ROOT]:
if not os.path.exists(root):
print(f"[WARNING] Directory not found: {root}")
continue
for folder in os.listdir(root):
if os.path.isdir(os.path.join(root, folder)):
identity_names.add(folder)
identity_names = sorted(list(identity_names))
total_identities = len(identity_names)
print(f"Total unique identities found: {total_identities}")
if total_identities == 0:
print("[ERROR] No identities found! Please check your dataset path.")
return
# 2. Split into train and test deterministically
split_idx = int(total_identities * SPLIT_RATIO)
train_ids = identity_names[:split_idx]
test_ids = identity_names[split_idx:]
print(f"Train identities (80%): {len(train_ids)}")
print(f"Test identities (20%): {len(test_ids)}")
# 3. Write to .txt files
train_txt_path = os.path.join(OUTPUT_DIR, "train_identities.txt")
test_txt_path = os.path.join(OUTPUT_DIR, "test_identities.txt")
os.makedirs(OUTPUT_DIR, exist_ok=True)
with open(train_txt_path, "w", encoding="utf-8") as f:
for identity in train_ids:
f.write(identity + "\n")
with open(test_txt_path, "w", encoding="utf-8") as f:
for identity in test_ids:
f.write(identity + "\n")
print(f"Saved train split to: {train_txt_path}")
print(f"Saved test split to: {test_txt_path}")
print("=" * 70)
if __name__ == "__main__":
main()