test.py

import pandas as pd
import string
import random

# ------------------------
# Config
# ------------------------
csv_file = "image_data.csv"
img_dir = "resized_plates"

# Allowed vocab (Tamil Nadu plates: TN00AA0000 format)
allowed_chars = set(string.ascii_uppercase + string.digits)  # A-Z, 0-9
special_tokens = {"<pad>", "<sos>", "<eos>"}

# ------------------------
# Load CSV
# ------------------------
df = pd.read_csv(csv_file)

if "license_plate" not in df.columns or "image_name" not in df.columns:
    raise ValueError("CSV must have columns: 'image_name', 'license_plate'")

print(f"✅ Loaded {len(df)} samples from {csv_file}")

# ------------------------
# Collect stats
# ------------------------
all_chars = set("".join(df["license_plate"].astype(str).tolist()))
invalid_chars = all_chars - allowed_chars

print("\n📊 Dataset Stats:")
print(f"Unique characters in labels: {sorted(all_chars)}")
print(f"Total unique characters: {len(all_chars)}")

if invalid_chars:
    print(f"\n⚠️ Found invalid characters: {sorted(invalid_chars)}")
else:
    print("\n✅ All characters are valid (A-Z, 0-9 only)")

# ------------------------
# Sample rows
# ------------------------
print("\n🔎 Sample entries:")
for i in random.sample(range(len(df)), min(10, len(df))):
    row = df.iloc[i]
    print(f"Image: {img_dir}/{row['image_name']} -> Label: {row['license_plate']}")