同步VSR中带来的改进:

支持不同任务使用不同选区
选区部分全部采用浮点值, 以便更好的适配不同尺寸不同分辨率
支持设置视频保存路径
将字幕选区改为实体对象
美化右键菜单样式
支持合并进度展示

Author: eritpchy

backend/bean/subtitle_area.py

@@ -0,0 +1,48 @@
+
+from typing import Union
+from dataclasses import dataclass
+from shapely.geometry import Polygon
+
+@dataclass
+class SubtitleArea:
+    """
+    字幕区域
+    """
+    ymin: Union[int, float]
+    ymax: Union[int, float]
+    xmin: Union[int, float]
+    xmax: Union[int, float]
+    # 字幕区域在视频中的位置
+    ab_section: range = None
+    
+    def __init__(self, ymin: Union[int, float], ymax: Union[int, float], 
+                 xmin: Union[int, float], xmax: Union[int, float], 
+                 ab_section: range = None):
+        self.ymin = ymin
+        self.ymax = ymax    
+        self.xmin = xmin
+        self.xmax = xmax
+        self.ab_section = ab_section
+
+    def normalized(self):
+        if self.xmin > self.xmax:
+            self.xmin, self.xmax = self.xmax, self.xmin
+        if self.ymin > self.ymax:
+            self.ymin, self.ymax = self.ymax, self.ymin
+
+    def is_empty(self):
+        return self.xmin == 0 and self.xmax == 0 and self.ymin == 0 and self.ymax == 0
+
+    @property
+    def width(self):
+        return self.xmax - self.xmin
+
+    @property
+    def height(self):
+        return self.ymax - self.ymin
+
+    def in_ab_section(self, frame_idx):
+        return True
+
+    def to_polygon(self):
+        return Polygon([[self.xmin, self.ymin], [self.xmax, self.ymin], [self.xmax, self.ymax], [self.xmin, self.ymax]])

backend/config.py

@@ -2,7 +2,7 @@
 import os
 from pathlib import Path
 from qfluentwidgets import (qconfig, ConfigItem, QConfig, OptionsValidator, BoolValidator, OptionsConfigItem, 
-                            EnumSerializer, RangeValidator, RangeConfigItem)
+                            EnumSerializer, RangeValidator, RangeConfigItem, ConfigValidator)
 from backend.tools.constant import SubtitleArea, VideoSubFinderDecoder
 import configparser
 
@@ -44,10 +44,9 @@ class Config(QConfig):
     windowW = ConfigItem("Window", "Width", 1200)
     windowH = ConfigItem("Window", "Height", 1200)
 
-    subtitleSelectionAreaX = ConfigItem("Main", "SubtitleSelectionAreaX", 0.05)
-    subtitleSelectionAreaY = ConfigItem("Main", "SubtitleSelectionAreaY", 0.78)
-    subtitleSelectionAreaW = ConfigItem("Main", "SubtitleSelectionAreaW", 0.90)
-    subtitleSelectionAreaH = ConfigItem("Main", "SubtitleSelectionAreaH", 0.21)
+    # 使用一个配置项存储所有选区
+    # 默认值为一个选区，格式为："ymin,ymax,xmin,xmax;ymin,ymax,xmin,xmax;..."，分号分隔不同选区
+    subtitleSelectionAreas = ConfigItem("Main", "SubtitleSelectionAreas", "0.78,0.99,0.05,0.95")
 
     # 字幕语言设置
     language = OptionsConfigItem("Main", "Language", "ch", OptionsValidator([name for name in tr["Language"]]))
@@ -91,6 +90,8 @@ class Config(QConfig):
     hardwareAcceleration = ConfigItem("Main", "HardwareAcceleration", HARDWARD_ACCELERATION_OPTION, BoolValidator())
     # 启动时检查应用更新
     checkUpdateOnStartup = ConfigItem("Main", "CheckUpdateOnStartup", True, BoolValidator())
+    # 视频保存目录
+    saveDirectory = ConfigItem("Main", "SaveDirectory", "", ConfigValidator())
     # VideoSubFinder CPU核心数
     videoSubFinderCpuCores = RangeConfigItem("Main", "VideoSubFinderCpuCores", 0, RangeValidator(0, os.cpu_count()))
     # VideoSubFinder 视频解码组件

backend/interface/ch.ini

@@ -55,6 +55,9 @@ VideoSubFinderCpuCores = CPU核心数
 VideoSubFinderCpuCoresDesc = 字幕提取时使用的CPU核心数，默认为0，即自动, 选择所有核心并不会变得快
 VideoSubFinderDecoder = 视频解码组件
 VideoSubFinderDecoderDesc = 用于针对不同视频兼容性需求, 默认为OpenCV(推荐), 当遇到无法正常提取时可以切换为FFmpeg(可能会造成时间轴细微偏移)
+ChooseDirectory = 选择文件夹
+SaveDirectory = 字幕保存目录
+SaveDirectoryDefault = 默认保存到输入视频当前目录
 
 [SubtitleArea]
 LowerPart = 下半部分
@@ -176,6 +179,7 @@ OpenVideoFirst = 请先打开视频
 SubtitleArea = 字幕区域
 VideoPreview = 视频预览
 ErrorDuringProcessing = 处理过程中发生错误: {}
+DeleteSelection = 删除当前激活选区
 
 [Main]
 RecSubLang = 识别字幕语言

backend/interface/chinese_cht.ini

@@ -55,6 +55,9 @@ VideoSubFinderCpuCores = CPU核心數
 VideoSubFinderCpuCoresDesc = 提取字幕時使用的CPU核心數，預設為0（自動），選擇所有核心不一定更快
 VideoSubFinderDecoder = 視頻解碼組件
 VideoSubFinderDecoderDesc = 用於不同影片相容性需求，預設為OpenCV（推薦），若無法正常提取可切換為FFmpeg（可能導致時間軸略有偏移）
+ChooseDirectory = 選擇資料夾
+SaveDirectory = 字幕儲存目錄
+SaveDirectoryDefault = 預設儲存至輸入影片的目前目錄
 
 [SubtitleArea]
 LowerPart = 下半部分
@@ -176,6 +179,7 @@ OpenVideoFirst = 請先打開視頻
 SubtitleArea = 字幕區域
 VideoPreview = 視頻預覽
 ErrorDuringProcessing = 處理過程中發生錯誤: {} 
+DeleteSelection = 刪除當前激活選區
 
 [Main]
 RecSubLang = 識別字幕語言

backend/interface/en.ini

@@ -55,6 +55,9 @@ VideoSubFinderCpuCores = CPU Cores
 VideoSubFinderCpuCoresDesc = Number of CPU cores used for subtitle extraction. Default is 0 (auto). Using all cores may not be faster.
 VideoSubFinderDecoder = Video Decoder
 VideoSubFinderDecoderDesc = For different video compatibility needs. Default is OpenCV (recommended). Switch to FFmpeg if extraction fails (may cause slight timeline shift).
+ChooseDirectory = Choose Folder
+SaveDirectory = Subtitle Save Directory
+SaveDirectoryDefault = Default: save to the current directory of the input video
 
 [SubtitleArea]
 LowerPart = Lower Part
@@ -177,6 +180,7 @@ OpenVideoFirst = Please Open Video First
 SubtitleArea = Subtitle Area
 VideoPreview = Video Preview
 ErrorDuringProcessing = Error during processing: {}
+DeleteSelection = Delete Selection
 
 [Main]
 RecSubLang = Subtitle Language

backend/interface/es.ini

@@ -55,6 +55,9 @@ VideoSubFinderCpuCores = Núcleos de CPU
 VideoSubFinderCpuCoresDesc = Núcleos de CPU usados para extraer subtítulos. Por defecto es 0 (automático). Usar todos los núcleos no siempre es más rápido.
 VideoSubFinderDecoder = Componente de decodificación de video
 VideoSubFinderDecoderDesc = Para diferentes necesidades de compatibilidad de video. Por defecto es OpenCV (recomendado). Cambia a FFmpeg si la extracción falla (puede causar un pequeño desfase en la línea de tiempo).
+ChooseDirectory = Seleccionar Carpeta
+SaveDirectory = Directorio de guardado de subtítulos
+SaveDirectoryDefault = Predeterminado: guardar en el directorio actual del video de entrada
 
 [SubtitleArea]
 LowerPart = Parte inferior
@@ -176,6 +179,7 @@ OpenVideoFirst = Por favor, abra el video primero
 SubtitleArea = Área de subtítulos
 VideoPreview = Vista previa de video
 ErrorDuringProcessing = Error durante el procesamiento: {}
+DeleteSelection = Eliminar selección
 
 [Main]
 RecSubLang = Idioma de subtítulos

backend/interface/japan.ini

@@ -55,6 +55,9 @@ VideoSubFinderCpuCores = CPUコア数
 VideoSubFinderCpuCoresDesc = 字幕抽出に使用するCPUコア数。デフォルトは0（自動）。全コアを選択しても速くなるとは限りません。
 VideoSubFinderDecoder = ビデオデコーダー
 VideoSubFinderDecoderDesc = 異なる動画互換性のニーズに対応。デフォルトはOpenCV（推奨）。抽出できない場合はFFmpegに切り替えてください（タイムラインがわずかにずれる場合があります）。
+ChooseDirectory = フォルダを選択
+SaveDirectory = 字幕の保存ディレクトリ
+SaveDirectoryDefault = デフォルト: 入力動画の現在のディレクトリに保存
 
 [SubtitleArea]
 LowerPart = 下部
@@ -176,6 +179,7 @@ OpenVideoFirst = 最初にビデオを開いてください
 SubtitleArea = サブタイトル領域
 VideoPreview = ビデオプレビュー
 ErrorDuringProcessing = 処理中にエラーが発生しました: {}
+DeleteSelection = 現在のアクティブ選択範囲を削除
 
 [Main]
 RecSubLang = サブタイトル言語

backend/interface/ko.ini

@@ -55,6 +55,9 @@ VideoSubFinderCpuCores = CPU 코어 수
 VideoSubFinderCpuCoresDesc = 자막 추출에 사용할 CPU 코어 수입니다. 기본값은 0(자동)이며, 모든 코어를 선택해도 더 빨라지지 않을 수 있습니다.
 VideoSubFinderDecoder = 비디오 디코더
 VideoSubFinderDecoderDesc = 다양한 비디오 호환성 요구에 사용. 기본값은 OpenCV(추천)이며, 추출이 실패할 경우 FFmpeg로 전환하세요(타임라인이 약간 어긋날 수 있음).
+ChooseDirectory = 폴더 선택
+SaveDirectory = 자막 저장 디렉터리
+SaveDirectoryDefault = 기본값: 입력 동영상의 현재 디렉터리에 저장
 
 [SubtitleArea]
 LowerPart = 하단 부분
@@ -176,6 +179,7 @@ OpenVideoFirst = 비디오을 먼저 열어주세요
 SubtitleArea = 자막 영역
 VideoPreview = 비디오 미리보기
 ErrorDuringProcessing = 처리 중 오류: {}
+DeleteSelection = 현재 활성 선택 영역 삭제
 
 [Main]
 RecSubLang = 자막 언어 인식

backend/interface/vi.ini

@@ -55,6 +55,10 @@ VideoSubFinderCpuCores = Số lõi CPU
 VideoSubFinderCpuCoresDesc = Số lõi CPU dùng để tách phụ đề. Mặc định là 0 (tự động). Dùng tất cả lõi không phải lúc nào cũng nhanh hơn.
 VideoSubFinderDecoder = Bộ giải mã video
 VideoSubFinderDecoderDesc = Dùng cho các nhu cầu tương thích video khác nhau. Mặc định là OpenCV (khuyến nghị). Chuyển sang FFmpeg nếu tách phụ đề thất bại (có thể lệch nhẹ thời gian).
+ChooseDirectory = Chọn Thư Mục
+SaveDirectory = Thư mục lưu phụ đề
+SaveDirectoryDefault = Mặc định: lưu vào thư mục hiện tại của video đầu vào
+DeleteSelection = Xóa lựa chọn
 
 [SubtitleArea]
 LowerPart = Phần dưới

backend/main.py

@@ -9,6 +9,7 @@
 import re
 import random
 import shutil
+import traceback
 from collections import Counter, namedtuple
 import unicodedata
 from threading import Thread
@@ -30,46 +31,24 @@
 from backend.tools import subtitle_ocr
 from backend.tools.paddle_model_config import PaddleModelConfig
 from backend.tools.process_manager import ProcessManager
+from backend.tools.subtitle_detect import SubtitleDetect
+from backend.bean.subtitle_area import SubtitleArea
 import threading
 import platform
 import multiprocessing
 import time
 import pysrt
 
-class SubtitleDetect:
-    """
-    文本框检测类，用于检测视频帧中是否存在文本框
-    """
-
-    def __init__(self):
-        from paddleocr.tools.infer import utility
-        from paddleocr.tools.infer.predict_det import TextDetector
-        hardware_accelerator = HardwareAccelerator.instance()
-        onnx_providers = hardware_accelerator.onnx_providers
-        model_config = PaddleModelConfig(hardware_accelerator)
-        args = utility.parse_args()
-        args.det_algorithm = 'DB'
-        args.det_model_dir = model_config.convertToOnnxModelIfNeeded(model_config.DET_MODEL_PATH)
-        args.use_gpu=hardware_accelerator.has_cuda()
-        args.use_onnx=len(onnx_providers) > 0
-        args.onnx_providers=onnx_providers
-        self.text_detector = TextDetector(args)
-
-    def detect_subtitle(self, img):
-        dt_boxes, elapse = self.text_detector(img)
-        return dt_boxes, elapse
-
-
 class SubtitleExtractor:
     """
     视频字幕提取类
     """
 
-    def __init__(self, vd_path, sub_area=None):
+    def __init__(self, vd_path):
         # 线程锁
         self.lock = threading.RLock()
         # 用户指定的字幕区域位置
-        self.sub_area = sub_area
+        self.sub_area = None
         self.hardware_accelerator = HardwareAccelerator.instance()
         # 是否使用硬件加速
         self.hardware_accelerator.set_enabled(config.hardwareAcceleration.value)
@@ -105,7 +84,7 @@ def __init__(self, vd_path, sub_area=None):
         # 自定义ocr对象
         self.ocr = None
         # 总处理进度
-        self.progress_total = 0
+        self.progress_total = 200
         # 视频帧提取进度
         self.progress_frame_extract = 0
         # OCR识别进度
@@ -231,12 +210,12 @@ def capture_frame_with_subtitle_area(self):
 
         if ret:
             # 如果有字幕区域，绘制矩形
-            if self.sub_area is not None:
-                s_ymin, s_ymax, s_xmin, s_xmax = self.sub_area
+            sub_area = self.sub_area
+            if sub_area is not None:
                 # 绘制绿色矩形框
-                cv2.rectangle(frame, (s_xmin, s_ymin), (s_xmax, s_ymax), (0, 255, 0), 2)
+                cv2.rectangle(frame, (sub_area.xmin, sub_area.ymin), (sub_area.xmax, sub_area.ymax), (0, 255, 0), 2)
                 # 添加文字标注
-                cv2.putText(frame, "Subtitle Area", (s_xmin, s_ymin - 10), 
+                cv2.putText(frame, "Subtitle Area", (sub_area.xmin, sub_area.ymin - 10), 
                            cv2.FONT_HERSHEY_SIMPLEX, 0.9, (0, 255, 0), 2)
 
             # 保存图像
@@ -303,15 +282,15 @@ def extract_frame_by_det(self):
             tbar.update(1)
             dt_boxes, elapse = self.sub_detector.detect_subtitle(frame)
             has_subtitle = False
-            if self.sub_area is not None:
-                s_ymin, s_ymax, s_xmin, s_xmax = self.sub_area
+            sub_area = self.sub_area
+            if sub_area is not None:
                 coordinate_list = get_coordinates(dt_boxes.tolist())
                 if coordinate_list:
                     for coordinate in coordinate_list:
                         xmin, xmax, ymin, ymax = coordinate
-                        if (s_xmin <= xmin and xmax <= s_xmax
-                                and s_ymin <= ymin
-                                and ymax <= s_ymax):
+                        if (sub_area.xmin <= xmin and xmax <= sub_area.xmax
+                                and sub_area.ymin <= ymin
+                                and ymax <= sub_area.ymax):
                             has_subtitle = True
                             # 检测到字幕时，如果列表为空，则为字幕头
                             if first_flag:
@@ -476,13 +455,13 @@ def vsf_output(out, ):
             path_vsf = os.path.join(BASE_DIR, 'subfinder', 'linux', 'VideoSubFinderCli.run')
             os.chmod(path_vsf, 0o775)
         # ：图像上半部分所占百分比，取值【0-1】
-        top_end = 1 - self.sub_area[0] / self.frame_height
+        top_end = 1 - self.sub_area.ymin / self.frame_height
         # bottom_end：图像下半部分所占百分比，取值【0-1】
-        bottom_end = 1 - self.sub_area[1] / self.frame_height
+        bottom_end = 1 - self.sub_area.ymax / self.frame_height
         # left_end：图像左半部分所占百分比，取值【0-1】
-        left_end = self.sub_area[2] / self.frame_width
+        left_end = self.sub_area.xmin / self.frame_width
         # re：图像右半部分所占百分比，取值【0-1】
-        right_end = self.sub_area[3] / self.frame_width
+        right_end = self.sub_area.xmax / self.frame_width
         if (not self.hardware_accelerator.has_cuda()) and len(self.hardware_accelerator.onnx_providers) > 0:
             cpu_count = multiprocessing.cpu_count()
         else:
@@ -932,16 +911,13 @@ def __get_area_text(self, ocr_result):
         coordinates = get_coordinates(box)
         area_text = []
         for content, coordinate in zip(text, coordinates):
-            if self.sub_area is not None:
-                s_ymin = self.sub_area[0]
-                s_ymax = self.sub_area[1]
-                s_xmin = self.sub_area[2]
-                s_xmax = self.sub_area[3]
+            sub_area = self.sub_area
+            if sub_area is not None:
                 xmin = coordinate[0]
                 xmax = coordinate[1]
                 ymin = coordinate[2]
                 ymax = coordinate[3]
-                if s_xmin <= xmin and xmax <= s_xmax and s_ymin <= ymin and ymax <= s_ymax:
+                if sub_area.xmin <= xmin and xmax <= sub_area.xmax and sub_area.ymin <= ymin and ymax <= sub_area.ymax:
                     area_text.append(content[0])
         return area_text
 
@@ -1019,10 +995,8 @@ def update_progress(self, ocr=None, frame_extract=None):
         if ocr is not None:
             self.progress_ocr = max(0, min(100, ocr))  # Clamp value between 0 and 100
         if frame_extract is not None:
-            self.progress_frame_extract = max(0, min(100, frame_extract))  # Clamp value between 0 and 100
-
-        self.progress_total = (self.progress_frame_extract * 0.4) + (self.progress_ocr * 0.6)
-        # Notify listeners
+            self.progress_frame_extract = max(0, min(100, frame_extract))
+        # 通知所有监听器
         self.notify_progress_listeners()
 
     def start_subtitle_ocr_async(self):
@@ -1080,7 +1054,7 @@ def add_progress_listener(self, listener):
         添加进度监听器
         
         Args:
-            listener: 一个回调函数，接收参数 (progress_total, isFinished)
+            listener: 一个回调函数，接收参数 (progress_ocr, progress_frame_extract, progress_total, isFinished)
         """
         if listener not in self.progress_listeners:
             self.progress_listeners.append(listener)
@@ -1101,9 +1075,9 @@ def notify_progress_listeners(self):
         """
         for listener in self.progress_listeners:
             try:
-                listener(self.progress_total, self.isFinished)
+                listener(self.progress_ocr, self.progress_frame_extract, self.progress_total, self.isFinished)
             except Exception as e:
-                print(f"通知进度监听器时出错: {str(e)}")
+                traceback.print_exc()
 
     def manage_process(pid):
         pass
@@ -1116,10 +1090,11 @@ def manage_process(pid):
     try:
         y_min, y_max, x_min, x_max = map(int, input(
             f"{tr['Main']['ChooseSubArea']} (ymin ymax xmin xmax)：").split())
-        subtitle_area = (y_min, y_max, x_min, x_max)
+        subtitle_area = SubtitleArea(y_min, y_max, x_min, x_max)
     except ValueError as e:
         subtitle_area = None
     # 新建字幕提取对象
-    se = SubtitleExtractor(video_path, subtitle_area)
+    se = SubtitleExtractor(video_path)
+    se.sub_area = subtitle_area
     # 开始提取字幕
     se.run()