File size: 9,048 Bytes
e53e238 |
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 |
"""
Subtitle Extractor Module
Extracts subtitles from videos using OCR and generates SRT files
"""
import cv2
import sys
import os
from pathlib import Path
from collections import defaultdict
# Add backend to path
sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
from backend.main import SubtitleDetect
class SubtitleExtractor:
"""Extract subtitles from video and generate SRT files"""
def __init__(self, video_path, sub_area=None):
"""
Initialize subtitle extractor
Args:
video_path: Path to video file
sub_area: Optional subtitle area (ymin, ymax, xmin, xmax)
"""
self.video_path = video_path
self.sub_area = sub_area
self.detector = SubtitleDetect(video_path, sub_area)
# Get video properties
self.video_cap = cv2.VideoCapture(video_path)
self.fps = self.video_cap.get(cv2.CAP_PROP_FPS)
self.frame_count = int(self.video_cap.get(cv2.CAP_PROP_FRAME_COUNT))
@property
def text_recognizer(self):
"""Lazy load PaddleOCR text recognizer"""
if not hasattr(self, '_text_recognizer'):
import paddle
paddle.disable_signal_handler()
from paddleocr.tools.infer import utility
from paddleocr.tools.infer.predict_rec import TextRecognizer
import importlib
import config
importlib.reload(config)
args = utility.parse_args()
args.rec_algorithm = 'CRNN'
args.rec_model_dir = config.REC_MODEL_PATH if hasattr(config, 'REC_MODEL_PATH') else os.path.join(config.DET_MODEL_BASE, config.MODEL_VERSION, 'ch_rec')
args.use_onnx = len(config.ONNX_PROVIDERS) > 0
args.onnx_providers = config.ONNX_PROVIDERS
self._text_recognizer = TextRecognizer(args)
return self._text_recognizer
def extract_text_from_frame(self, frame, boxes):
"""
Extract text from frame using OCR
Args:
frame: Video frame (numpy array)
boxes: List of detected text boxes [(xmin, xmax, ymin, ymax), ...]
Returns:
List of extracted text strings
"""
texts = []
for box in boxes:
xmin, xmax, ymin, ymax = box
# Crop text region
text_region = frame[ymin:ymax, xmin:xmax]
if text_region.size == 0:
continue
try:
# Run OCR on cropped region
rec_result, _ = self.text_recognizer([text_region])
if rec_result and len(rec_result) > 0:
text, confidence = rec_result[0]
if confidence > 0.5: # Only accept if confidence > 50%
texts.append(text)
except Exception as e:
print(f"Warning: OCR failed for box {box}: {e}")
continue
return texts
def format_timestamp(self, seconds):
"""
Convert seconds to SRT timestamp format (HH:MM:SS,mmm)
Args:
seconds: Time in seconds (float)
Returns:
Formatted timestamp string
"""
hours = int(seconds // 3600)
minutes = int((seconds % 3600) // 60)
secs = int(seconds % 60)
millis = int((seconds % 1) * 1000)
return f"{hours:02d}:{minutes:02d}:{secs:02d},{millis:03d}"
def extract_subtitles(self, progress_callback=None):
"""
Extract subtitles with OCR and timestamps
Args:
progress_callback: Optional callback function for progress updates
Returns:
List of subtitle dictionaries with 'start', 'end', 'text' keys
"""
print("[Subtitle Extractor] Starting subtitle extraction...")
# Detect subtitle regions
subtitle_frame_dict = self.detector.find_subtitle_frame_no()
if not subtitle_frame_dict:
print("[Subtitle Extractor] No subtitles detected!")
return []
print(f"[Subtitle Extractor] Found subtitles in {len(subtitle_frame_dict)} frames")
# Group continuous frames with same text
subtitles = []
current_subtitle = None
# Reset video capture
self.video_cap.set(cv2.CAP_PROP_POS_FRAMES, 0)
current_frame_no = 0
# Find continuous ranges
continuous_ranges = self.detector.find_continuous_ranges_with_same_mask(subtitle_frame_dict)
for start_frame, end_frame in continuous_ranges:
# Seek to start frame
self.video_cap.set(cv2.CAP_PROP_POS_FRAMES, start_frame - 1)
ret, frame = self.video_cap.read()
if not ret:
continue
# Get boxes for this frame
boxes = subtitle_frame_dict.get(start_frame, [])
# Extract text
texts = self.extract_text_from_frame(frame, boxes)
combined_text = " ".join(texts).strip()
if not combined_text:
continue
# Calculate timestamps
start_time = (start_frame - 1) / self.fps
end_time = end_frame / self.fps
# Merge with previous if same text and continuous
if (current_subtitle and
current_subtitle['text'] == combined_text and
abs(start_time - current_subtitle['end']) < 1.0):
# Extend end time
current_subtitle['end'] = end_time
else:
# Add previous subtitle if exists
if current_subtitle:
subtitles.append(current_subtitle)
# Start new subtitle
current_subtitle = {
'start': start_time,
'end': end_time,
'text': combined_text
}
if progress_callback:
progress = end_frame / self.frame_count
progress_callback(progress, f"Extracting subtitles: {len(subtitles)+1} found")
# Add last subtitle
if current_subtitle:
subtitles.append(current_subtitle)
print(f"[Subtitle Extractor] Extracted {len(subtitles)} subtitle segments")
return subtitles
def generate_srt(self, subtitles, output_path):
"""
Generate SRT file from subtitles
Args:
subtitles: List of subtitle dictionaries
output_path: Path to save SRT file
Returns:
Path to generated SRT file
"""
print(f"[Subtitle Extractor] Generating SRT file: {output_path}")
with open(output_path, 'w', encoding='utf-8') as f:
for i, sub in enumerate(subtitles, 1):
# Subtitle number
f.write(f"{i}\n")
# Timestamps
start_ts = self.format_timestamp(sub['start'])
end_ts = self.format_timestamp(sub['end'])
f.write(f"{start_ts} --> {end_ts}\n")
# Text
f.write(f"{sub['text']}\n")
# Blank line
f.write("\n")
print(f"[Subtitle Extractor] SRT file saved: {output_path}")
return output_path
def extract_to_srt(self, output_path=None, progress_callback=None):
"""
Complete extraction pipeline: detect -> OCR -> generate SRT
Args:
output_path: Optional custom output path for SRT file
progress_callback: Optional callback for progress updates
Returns:
Path to generated SRT file
"""
# Default output path
if output_path is None:
video_name = Path(self.video_path).stem
output_dir = Path(self.video_path).parent
output_path = output_dir / f"{video_name}_subtitles.srt"
# Extract subtitles
subtitles = self.extract_subtitles(progress_callback)
if not subtitles:
# Create empty SRT
with open(output_path, 'w', encoding='utf-8') as f:
f.write("# No subtitles detected\n")
return str(output_path)
# Generate SRT
return self.generate_srt(subtitles, str(output_path))
if __name__ == '__main__':
import sys
if len(sys.argv) < 2:
print("Usage: python subtitle_extractor.py <video_path>")
sys.exit(1)
video_path = sys.argv[1]
extractor = SubtitleExtractor(video_path)
srt_path = extractor.extract_to_srt()
print(f"Subtitles extracted to: {srt_path}")
|