122 lines
4.1 KiB
Python
122 lines
4.1 KiB
Python
#!/usr/bin/env python
|
|
# -*- coding: utf-8 -*-
|
|
"""
|
|
Curved Seal Text Extraction
|
|
1. First crop the seal region from sanity_check.png
|
|
2. Use SealTextDetection to find curved text regions
|
|
3. Use TextRecognition for each region
|
|
"""
|
|
import sys
|
|
import json
|
|
import os
|
|
from PIL import Image, ImageDraw, ImageFont
|
|
import numpy as np
|
|
|
|
def main():
|
|
if len(sys.argv) < 2:
|
|
print(json.dumps({"error": "Usage: python seal_curved_ocr.py <image_path>"}))
|
|
sys.exit(1)
|
|
|
|
image_path = sys.argv[1]
|
|
|
|
try:
|
|
from paddleocr import LayoutDetection, SealTextDetection, TextRecognition
|
|
|
|
print("Step 1: Detecting seal region...", file=sys.stderr)
|
|
|
|
# Step 1: Find seal in the image
|
|
layout_model = LayoutDetection(model_name="PP-DocLayout_plus-L")
|
|
layout_output = layout_model.predict(image_path, batch_size=1, layout_nms=True)
|
|
|
|
seal_box = None
|
|
for res in layout_output:
|
|
for box in res["boxes"]:
|
|
if box["label"] == "seal":
|
|
seal_box = box["coordinate"]
|
|
break
|
|
|
|
if seal_box is None:
|
|
print(json.dumps({"error": "No seal found in image"}))
|
|
sys.exit(1)
|
|
|
|
print(f"Found seal at: {seal_box}", file=sys.stderr)
|
|
|
|
# Step 2: Crop seal region
|
|
img = Image.open(image_path)
|
|
x1, y1, x2, y2 = [int(c) for c in seal_box]
|
|
padding = 20
|
|
x1 = max(0, x1 - padding)
|
|
y1 = max(0, y1 - padding)
|
|
x2 = min(img.width, x2 + padding)
|
|
y2 = min(img.height, y2 + padding)
|
|
|
|
seal_crop = img.crop((x1, y1, x2, y2))
|
|
seal_crop_path = "temp_seal_crop.png"
|
|
seal_crop.save(seal_crop_path)
|
|
|
|
print("Step 2: Detecting text in seal...", file=sys.stderr)
|
|
|
|
# Step 3: Detect text in seal
|
|
seal_det = SealTextDetection(model_name="PP-OCRv4_server_seal_det")
|
|
det_output = seal_det.predict(seal_crop_path, batch_size=1)
|
|
|
|
# Step 4: Recognize text
|
|
text_rec = TextRecognition(model_name="PP-OCRv4_server_rec")
|
|
|
|
all_texts = []
|
|
draw = ImageDraw.Draw(seal_crop)
|
|
|
|
for det_res in det_output:
|
|
dt_polys = det_res.get("dt_polys", [])
|
|
print(f"Found {len(dt_polys)} text regions", file=sys.stderr)
|
|
|
|
for poly in dt_polys:
|
|
poly = np.array(poly)
|
|
# Get bounding box
|
|
min_x, min_y = poly.min(axis=0)
|
|
max_x, max_y = poly.max(axis=0)
|
|
|
|
# Crop text region
|
|
text_crop = seal_crop.crop((int(min_x), int(min_y), int(max_x), int(max_y)))
|
|
text_crop_path = "temp_text_crop.png"
|
|
text_crop.save(text_crop_path)
|
|
|
|
# Recognize
|
|
rec_output = text_rec.predict(text_crop_path, batch_size=1)
|
|
for rec_res in rec_output:
|
|
text = rec_res.get("rec_text", "")
|
|
score = rec_res.get("rec_score", 0)
|
|
if hasattr(score, 'item'):
|
|
score = score.item()
|
|
|
|
all_texts.append({
|
|
"text": str(text),
|
|
"score": float(score)
|
|
})
|
|
print(f" - Recognized: '{text}' (score: {score:.2f})", file=sys.stderr)
|
|
|
|
# Draw polygon
|
|
pts = [(int(p[0]), int(p[1])) for p in poly]
|
|
draw.polygon(pts, outline=(255, 0, 255), width=2)
|
|
|
|
os.remove(text_crop_path)
|
|
|
|
# Save annotated seal
|
|
seal_crop.save("seal_annotated.png")
|
|
os.remove(seal_crop_path)
|
|
|
|
combined = " ".join([t["text"] for t in all_texts])
|
|
print(json.dumps({
|
|
"success": True,
|
|
"texts": all_texts,
|
|
"combined_text": combined
|
|
}, ensure_ascii=False))
|
|
|
|
except Exception as e:
|
|
import traceback
|
|
print(json.dumps({"error": str(e), "traceback": traceback.format_exc()}))
|
|
sys.exit(1)
|
|
|
|
if __name__ == "__main__":
|
|
main()
|