| |
|
| | import os
|
| | import sys
|
| | import argparse
|
| | import json
|
| | import datetime
|
| | import cv2
|
| | import numpy as np
|
| | import torch
|
| | import lpips
|
| | from torchvision import transforms
|
| | import torch.nn.functional as F
|
| | from PIL import Image, UnidentifiedImageError
|
| |
|
| | def verify_image(path, exts=('.png', '.jpg', '.jpeg', '.webp')):
|
| | """Verify file exists, is non-empty, has valid extension, and can be opened by PIL."""
|
| | if not os.path.isfile(path):
|
| | return False, f'File not found: {path}'
|
| | if os.path.getsize(path) == 0:
|
| | return False, f'File is empty: {path}'
|
| | if not path.lower().endswith(exts):
|
| | return False, f'Unsupported format: {path}'
|
| | try:
|
| | img = Image.open(path)
|
| | img.verify()
|
| | except (UnidentifiedImageError, Exception) as e:
|
| | return False, f'Cannot read image: {path} ({e})'
|
| | return True, ''
|
| |
|
| | def load_tensor(path):
|
| | """Load and normalize image to [-1,1] range Tensor"""
|
| | img = cv2.imread(path, cv2.IMREAD_COLOR)
|
| | if img is None:
|
| | raise RuntimeError(f'cv2 read failed: {path}')
|
| | img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
|
| | t = transforms.ToTensor()(img) * 2 - 1
|
| | return t.unsqueeze(0)
|
| |
|
| | def histogram_intersection(a, b, bins=256):
|
| | """Calculate average intersection ratio of RGB channel histograms between two images"""
|
| | inters = []
|
| | for ch in range(3):
|
| | h1 = cv2.calcHist([a], [ch], None, [bins], [0,256]).ravel()
|
| | h2 = cv2.calcHist([b], [ch], None, [bins], [0,256]).ravel()
|
| | if h1.sum() > 0:
|
| | h1 = h1 / h1.sum()
|
| | if h2.sum() > 0:
|
| | h2 = h2 / h2.sum()
|
| | inters.append(np.minimum(h1, h2).sum())
|
| | return float(np.mean(inters))
|
| |
|
| | if __name__ == "__main__":
|
| | p = argparse.ArgumentParser(description='Automated style transfer evaluation script')
|
| | p.add_argument('--groundtruth', required=True, help='Ground truth directory path')
|
| | p.add_argument('--output', required=True, help='Path to stylized output image')
|
| | p.add_argument('--lpips-thresh', type=float, default=0.5, help='LPIPS threshold (>= passes)')
|
| | p.add_argument('--hi-thresh', type=float, default=0.7, help='HI (Histogram Intersection) threshold (>= passes)')
|
| | p.add_argument('--result', required=True, help='Result JSONL file path (append mode)')
|
| | args = p.parse_args()
|
| |
|
| |
|
| | content_path = os.path.join(args.groundtruth, 'gt_01.jpg')
|
| | style_path = os.path.join(args.groundtruth, 'gt_02.jpg')
|
| |
|
| | process = True
|
| | comments = []
|
| |
|
| |
|
| | for tag, path in [('content', content_path), ('style', style_path), ('output', args.output)]:
|
| | ok, msg = verify_image(path)
|
| | if not ok:
|
| | process = False
|
| | comments.append(f'[{tag}] {msg}')
|
| |
|
| |
|
| | lpips_pass = hi_pass = False
|
| | lpips_val = hi_val = None
|
| | if process:
|
| | try:
|
| |
|
| | img_c = load_tensor(content_path)
|
| | img_o = load_tensor(args.output)
|
| |
|
| | _, _, h0, w0 = img_c.shape
|
| | _, _, h1, w1 = img_o.shape
|
| | nh, nw = min(h0,h1), min(w0,w1)
|
| | if (h0,w0)!=(nh,nw):
|
| | img_c = F.interpolate(img_c, size=(nh,nw), mode='bilinear', align_corners=False)
|
| | if (h1,w1)!=(nh,nw):
|
| | img_o = F.interpolate(img_o, size=(nh,nw), mode='bilinear', align_corners=False)
|
| |
|
| | loss_fn = lpips.LPIPS(net='vgg').to(torch.device('cpu'))
|
| | with torch.no_grad():
|
| | lpips_val = float(loss_fn(img_c, img_o).item())
|
| | lpips_pass = lpips_val >= args.lpips_thresh
|
| |
|
| |
|
| | img_s = cv2.imread(style_path, cv2.IMREAD_COLOR)
|
| | img_o_cv = cv2.imread(args.output, cv2.IMREAD_COLOR)
|
| | hi_val = histogram_intersection(img_s, img_o_cv)
|
| | hi_pass = hi_val >= args.hi_thresh
|
| |
|
| | comments.append(f'LPIPS={lpips_val:.4f} (>= {args.lpips_thresh} → {"OK" if lpips_pass else "FAIL"})')
|
| | comments.append(f'HI={hi_val:.4f} (>= {args.hi_thresh} → {"OK" if hi_pass else "FAIL"})')
|
| |
|
| | except Exception as e:
|
| | process = False
|
| | comments.append(f'Metric calculation error: {e}')
|
| |
|
| |
|
| | result_flag = (process and lpips_pass and hi_pass)
|
| | entry = {
|
| | "Process": process,
|
| | "Result": result_flag,
|
| | "TimePoint": datetime.datetime.now().isoformat(sep='T', timespec='seconds'),
|
| | "comments": "; ".join(comments)
|
| | }
|
| | print(entry["comments"])
|
| | os.makedirs(os.path.dirname(args.result) or '.', exist_ok=True)
|
| | with open(args.result, 'a', encoding='utf-8') as f:
|
| | f.write(json.dumps(entry, ensure_ascii=False, default=str) + "\n")
|
| |
|
| |
|
| | print("\nEvaluation complete - Final status: " + ("PASSED" if result_flag else "FAILED")) |
