Spaces:
Running
Running
File size: 11,356 Bytes
47dc635 fc4abe8 47dc635 fc4abe8 47dc635 fc4abe8 47dc635 fc4abe8 47dc635 fc4abe8 47dc635 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 | from services.bubble_detector_kiuyha_service import Bubble_Detector_Kiuyha_Service
from services.translate_qwen_service import Translate_Qwen_Service
from PIL import Image, ImageDraw, ImageFont
import tempfile
import os
import re
import torch
from pathlib import Path
from helpers import get_project_root, setup_fonts
from manga_ocr import MangaOcr
import httpx
class ImageProcessor:
def __init__(self, bubble_detector, ocr_model, translate_model):
self.bubble_detector_model = bubble_detector
self.ocr_model = ocr_model
self.translate_model = translate_model
async def download_and_process(self, image_url: str, language: str):
# Create a temporary file that stays on disk until we close it
# 'delete=False' is important because some ML models need the file to stay closed/flushed before they can read it.
with tempfile.NamedTemporaryFile(suffix=".jpg", delete=False) as tmp:
temp_path = tmp.name
# Download
async with httpx.AsyncClient() as client:
response = await client.get(image_url)
response.raise_for_status()
tmp.write(response.content)
tmp.flush()
try:
results = self.process_image(temp_path, language)
return results
finally:
if os.path.exists(temp_path):
os.remove(temp_path)
print(f"Cleaned up temp file: {temp_path}")
def process_image(self, image_path, language):
bubble_results = self.bubble_detector_model.predict(image_path)
print(f"bubble results: {bubble_results}")
img = Image.open(image_path)
width, height = img.size
# draw = ImageDraw.Draw(img)
texts = []
coordinates={}
i=0
for box_data in bubble_results:
coords = box_data['coords']
# draw.rectangle(coords, outline="red", width=1)
box_cropped = img.crop(coords)
# box_cropped = upscale_for_ocr(box_cropped, scale=3)
# box_cropped.show()
with tempfile.NamedTemporaryFile(delete=False, suffix=".png") as f:
box_cropped.save(f.name)
temp_path = f.name
text = ""
try:
text = self.ocr_model(box_cropped)
except Exception as e:
print(f"text OCR failed for {i}")
text = re.sub(r'[\n\r\u2028\u2029]+', ' ', text) #remove new lines
texts.append({"id": i, "text": text})
coordinates[i] = coords
i+=1
print(f'OCR Complete, total {len(texts)} bubbles.')
#add translated text to manga image
try:
print("Translating with cloud Qwen model...")
translated = self.translate_model.translate_cloud(texts)
except Exception as e:
print("API translation failed with Qwen, falling back to local model...")
translated = self.translate_model.translate(texts)
print(translated)
bubble_data = []
for i in range(len(texts)):
coords = coordinates[i]
x1, y1, x2, y2 = coords
original_text = texts[i]["text"]
translated_text = translated.get(str(i), translated.get(i, ""))
if not isinstance(translated_text, str):
translated_text = str(translated_text)
print(f"{i}: {original_text}")
print(translated_text)
print("==================================")
bubble_data.append({
"bubble_index": i,
"width": width,
"height": height,
"x1": float(x1), "y1": float(y1), "x2": float(x2), "y2": float(y2),
"original_text": original_text,
"translated_text": translated_text,
})
######### Code for drawing translated text onto manga panel directly) ###########
# #wipe the space
# draw.rectangle(coords, fill="white", outline="white")
# # 1. Calculate the best fit
# lines, best_font, final_size, line_h = fit_text_to_box(draw, translated_text, coords, FONT_PATH)
# # Calculate total height of the block
# total_h = line_h * len(lines)
# # Start_y adjusted for the block height relative to the box center
# start_y = coords[1] + ((coords[3] - coords[1]) - total_h) / 2
# # 3. Draw each line centered horizontally
# for line in lines:
# line = line.strip()
# if not line: continue
# # Horizontal Centering
# line_w = draw.textlength(line, font=best_font)
# start_x = coords[0] + ((coords[2] - coords[0]) - line_w) / 2
# draw.text((start_x, start_y), line, font=best_font, fill="black")
# start_y += line_h
return bubble_data #img, bubble_data
########Test code, keeping it here as reference. Remove later################
# def show_boxes(image_path):
# result = bubble_detector_model.predict(image_path)
# img = Image.open(image_path).convert("RGB")
# draw = ImageDraw.Draw(img)
# for box in result.boxes:
# # Get coordinates as a list of floats
# coords = box.xyxy[0].tolist() # [x1, y1, x2, y2]
# draw.rectangle(coords, outline="red", width=1)
# # label
# conf = box.conf[0].item()
# box_cropped = img.crop(coords)
# # box_cropped = upscale_for_ocr(box_cropped, scale=3)
# with tempfile.NamedTemporaryFile(delete=False, suffix=".png") as f:
# box_cropped.save(f.name)
# temp_path = f.name
# draw.text(
# (coords[0], coords[1] - 10),
# "b",
# fill="red",
# font=font
# )
# img.show()
# def get_wrapped_text(text, font, max_width):
# lines = []
# words = text.split(' ') # Split by words for English
# current_line = []
# for word in words:
# # Check if adding the next word exceeds the width
# test_line = ' '.join(current_line + [word])
# # getlength() is more accurate than getbbox for text width
# if font.getlength(test_line) <= max_width:
# current_line.append(word)
# else:
# lines.append(' '.join(current_line))
# current_line = [word]
# lines.append(' '.join(current_line))
# return lines
# def fit_text_to_box(draw, text, box_coords, font_path, padding=5, initial_size=40):
# x1, y1, x2, y2 = box_coords
# padding = padding
# target_width = (x2 - x1) - (padding * 2)
# target_height = (y2 - y1) - (padding * 2)
# current_size = initial_size
# lines = []
# while current_size > 8:
# # index=0 for Japanese, 1 for Korean in NotoSansCJK
# font = ImageFont.truetype(font_path, size=current_size)
# lines = get_wrapped_text(text, font, target_width)
# # Use a more reliable line height measurement
# # getbbox can be inconsistent; use font.size * constant for better leading
# line_height = int(current_size * 1.2)
# total_height = line_height * len(lines)
# if total_height <= target_height:
# break
# current_size -= 2 # Step down by 2 for speed
# return lines, font, current_size, line_height
# def upscale_for_ocr(img, scale=2):
# w, h = img.size
# return img.resize((w*scale, h*scale), Image.BICUBIC)
# def process_image(image_path, language):
# bubble_results = bubble_detector_model.predict(image_path)
# print(f"bubble results: {bubble_results}")
# img = Image.open(image_path)
# draw = ImageDraw.Draw(img)
# texts = []
# coordinates={}
# i=0
# for box_data in bubble_results:
# coords = box_data['coords']
# draw.rectangle(coords, outline="red", width=1)
# box_cropped = img.crop(coords)
# # box_cropped = upscale_for_ocr(box_cropped, scale=3)
# # box_cropped.show()
# with tempfile.NamedTemporaryFile(delete=False, suffix=".png") as f:
# box_cropped.save(f.name)
# temp_path = f.name
# text = ""
# # if language == "japanese":
# # # text = ocr_japanese_model.runOCR(temp_path)
# # text = ocr_model(temp_path)
# # else:
# # text = ocr_model.runOCR(temp_path)
# text = ocr_model(box_cropped)
# text = re.sub(r'[\n\r\u2028\u2029]+', ' ', text) #remove new lines
# texts.append({"id": i, "text": text})
# coordinates[i] = coords
# i+=1
# print(f'OCR Complete, total {len(texts)} bubbles.')
# #add translated text to manga image
# try:
# print("Translating with cloud Qwen model...")
# translated = translate_model.translate_cloud(texts)
# except Exception as e:
# print("API translation failed with Qwen, falling back to local model...")
# translated = translate_model.translate(texts)
# print(translated)
# bubble_data = []
# for i in range(len(texts)):
# coords = coordinates[i]
# x1, y1, x2, y2 = coords
# original_text = texts[i]["text"]
# translated_text = translated.get(str(i), translated.get(i, ""))
# if not isinstance(translated_text, str):
# translated_text = str(translated_text)
# print(f"{i}: {original_text}")
# print(translated_text)
# print("==================================")
# bubble_data.append({
# "bubble_index": i,
# "x1": float(x1), "y1": float(y1), "x2": float(x2), "y2": float(y2),
# "original_text": original_text,
# "translated_text": translated_text,
# })
# #wipe the space
# draw.rectangle(coords, fill="white", outline="white")
# # 1. Calculate the best fit
# lines, best_font, final_size, line_h = fit_text_to_box(draw, translated_text, coords, FONT_PATH)
# # Calculate total height of the block
# total_h = line_h * len(lines)
# # Start_y adjusted for the block height relative to the box center
# start_y = coords[1] + ((coords[3] - coords[1]) - total_h) / 2
# # 3. Draw each line centered horizontally
# for line in lines:
# line = line.strip()
# if not line: continue
# # Horizontal Centering
# line_w = draw.textlength(line, font=best_font)
# start_x = coords[0] + ((coords[2] - coords[0]) - line_w) / 2
# draw.text((start_x, start_y), line, font=best_font, fill="black")
# start_y += line_h
# return img, bubble_data
# def translate_text(text, language):
# # translated_text = ""
# # if language == "japanese":
# # translated_text =
# translated_text = translate_model.translate(text)
# return translated_text
# def _language_to_code(language: str) -> str:
# """Map language name to ISO 639-1 style code for DB."""
# m = {"japanese": "ja", "english": "en", "korean": "ko", "chinese": "zh"}
# return m.get(language.lower(), language[:2] if len(language) >= 2 else "ja")
|