compare_decoders.py

import cv2
import os
import re
import time
from collections import defaultdict

BARCODE_DIR = "images/barcode"
PART_NUMBER_RE = re.compile(r'^0301BAB\d+N$')


def get_image_files():
    files = sorted([f for f in os.listdir(BARCODE_DIR)
                    if f.lower().endswith(('.jpg', '.jpeg', '.png'))])
    return files


def is_chassis(text):
    text = text.strip()
    return bool(text) and not PART_NUMBER_RE.match(text)




def decode_pyzbar(img, gray, thresh):
    from pyzbar.pyzbar import decode
    all_values = set()
    for frame in [img, thresh]:
        for d in decode(frame):
            text = d.data.decode("utf-8").strip()
            all_values.add(text)
    return all_values


def decode_zxingcpp(img, gray, thresh):
    import zxingcpp
    all_values = set()
    for frame in [gray, thresh]:
        try:
            results = zxingcpp.read_barcodes(frame)
            for r in results:
                text = r.text.strip()
                if text:
                    all_values.add(text)
        except Exception as e:
            pass
    return all_values


def decode_cv2barcode(img, gray, thresh):
    all_values = set()
    try:
        detector = cv2.barcode.BarcodeDetector()
        for frame in [gray, thresh]:
            ok, decoded_info, decoded_type, points = detector.detectAndDecode(frame)
            if ok and decoded_info is not None:
                for text in decoded_info:
                    if text and text.strip():
                        all_values.add(text.strip())
    except AttributeError:
        pass
    except Exception as e:
        pass
    return all_values


def main():
    files = get_image_files()
    print(f"Testing {len(files)} barcode images from {BARCODE_DIR}/\n")

    decoders = {
        "pyzbar":     decode_pyzbar,
        "zxing-cpp":  decode_zxingcpp,
        "cv2.barcode": decode_cv2barcode,
    }

    stats = {name: {"chassis": 0, "part_only": 0, "none": 0, "chassis_list": []}
             for name in decoders}
    timings = {name: 0.0 for name in decoders}

    detail_rows = []

    for fname in files:
        path = os.path.join(BARCODE_DIR, fname)
        img = cv2.imread(path)
        if img is None:
            continue
        gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
        _, thresh = cv2.threshold(gray, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)

        key = os.path.splitext(fname)[0]
        row = {"file": fname}

        for name, decoder_fn in decoders.items():
            t0 = time.perf_counter()
            all_values = decoder_fn(img, gray, thresh)
            elapsed = time.perf_counter() - t0
            timings[name] += elapsed

            chassis_vals = {v for v in all_values if is_chassis(v)}
            part_vals    = {v for v in all_values if PART_NUMBER_RE.match(v)}

            if chassis_vals:
                stats[name]["chassis"] += 1
                stats[name]["chassis_list"].append((key, chassis_vals))
                row[name] = ", ".join(sorted(chassis_vals))
            elif part_vals:
                stats[name]["part_only"] += 1
                row[name] = "(part# only)"
            else:
                stats[name]["none"] += 1
                row[name] = "—"

        detail_rows.append(row)

    print("=" * 100)
    print(f"{'File':<14} {'pyzbar':<20} {'zxing-cpp':<20} {'cv2.barcode':<20}")
    print("-" * 100)
    for row in detail_rows:
        pyz = row.get("pyzbar", "—")
        zxc = row.get("zxing-cpp", "—")
        cv2b = row.get("cv2.barcode", "—")
        print(f"{row['file']:<14} {pyz:<20} {zxc:<20} {cv2b:<20}")

    total = len(files)
    print("\n" + "=" * 100)
    print("SUMMARY")
    print("=" * 100)
    print(f"{'Metric':<30} {'pyzbar':>12} {'zxing-cpp':>12} {'cv2.barcode':>12}")
    print("-" * 70)
    print(f"{'Chassis decoded':.<30} {stats['pyzbar']['chassis']:>12} {stats['zxing-cpp']['chassis']:>12} {stats['cv2.barcode']['chassis']:>12}")
    print(f"{'Part# only (filtered out)':.<30} {stats['pyzbar']['part_only']:>12} {stats['zxing-cpp']['part_only']:>12} {stats['cv2.barcode']['part_only']:>12}")
    print(f"{'Nothing decoded':.<30} {stats['pyzbar']['none']:>12} {stats['zxing-cpp']['none']:>12} {stats['cv2.barcode']['none']:>12}")
    print(f"{'Total time (s)':.<30} {timings['pyzbar']:>12.2f} {timings['zxing-cpp']:>12.2f} {timings['cv2.barcode']:>12.2f}")
    print("-" * 70)
    print(f"{'CHASSIS DECODE RATE':.<30} {stats['pyzbar']['chassis']/total:>11.0%}  {stats['zxing-cpp']['chassis']/total:>11.0%}  {stats['cv2.barcode']['chassis']/total:>11.0%}")
    print("=" * 100)

    best_name = max(decoders.keys(), key=lambda n: stats[n]["chassis"])
    print(f"\n★ Best decoder: {best_name} ({stats[best_name]['chassis']}/{total} chassis barcodes)")

    for name in decoders:
        if stats[name]["chassis_list"]:
            print(f"\n  {name} decoded chassis numbers:")
            for key, vals in stats[name]["chassis_list"]:
                print(f"    {key}: {', '.join(sorted(vals))}")


if __name__ == "__main__":
    main()