count_samples.py

import os
import numpy as np

base_dir = r"c:\Users\ASUS\lung_ai_project\data"

paths = [
    os.path.join(base_dir, "hear_embeddings_optimized", "y_hear_opt_merged.npy"),
    os.path.join(base_dir, "hear_embeddings_coughvid", "y_coughvid.npy"),
    os.path.join(base_dir, "hear_embeddings_elite", "y_elite.npy"),
    os.path.join(base_dir, "hear_embeddings_augmented_new", "y_hear_aug_merged.npy")
]

total = 0
for p in paths:
    if os.path.exists(p):
        y = np.load(p)
        print(f"{os.path.basename(p)}: {len(y)} samples")
        total += len(y)
    else:
        print(f"MISSING: {p}")

print(f"\nGRAND TOTAL SAMPLES: {total}")