利用pyscenedetect实现类似剪映智能镜头分割

rootcup

PySceneDetect是一款基于opencv的视频场景切换检测和分析工具。本文代码也是利用pyscenedetect实现类似的功能

剪映的智能镜头分割


达芬奇的场景切割


均能实现自动识别镜头切换，然后分割


文末也附上打包好的exe程序
打包后如下，由于需要用到ffmpeg，顺带在根目录放一个ffmpeg.exe方便程序调用



图形界面如下



可以选择单个视频文件，也可选择视频文件夹批量处理

[Python] 纯文本查看 复制代码

?

001 002 003 004 005 006 007 008 009 010 011 012 013 014 015 016 017 018 019 020 021 022 023 024 025 026 027 028 029 030 031 032 033 034 035 036 037 038 039 040 041 042 043 044 045 046 047 048 049 050 051 052 053 054 055 056 057 058 059 060 061 062 063 064 065 066 067 068 069 070 071 072 073 074 075 076 077 078 079 080 081 082 083 084 085 086 087 088 089 090 091 092 093 094 095 096 097 098 099 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295

import base64 import concurrent.futures import io import json import os import subprocess import threading import tkinter as tk from ctypes import windll from tkinter import filedialog, messagebox, Toplevel   from PIL import Image, ImageTk from scenedetect import open_video, SceneManager from scenedetect.detectors import ContentDetector from scenedetect.video_splitter import split_video_ffmpeg   # 设置应用程序的默认DPI，确保高DPI屏幕上的显示效果 windll.shcore.SetProcessDpiAwareness(2)   font = 'simhei.ttf'  # 设置字体 win自带的，不用绝对路径 config_file = 'config.json'   # 多语言支持 languages = {     'zh': {         'title': '分镜探测切割 1.4',         'about': '关于',         'select_video_file': '选择视频文件',         'select_video_folder': '选择视频文件夹',         'threshold': '画面变化阈值:',         'start': '开始切割分镜',         'open_folder': '打开输出文件夹',         'readme': "2024年9月22日 by 吾爱破解：rootcup\n说明：画面变化阈值 数字越小越灵敏，切割的分镜头就越多。\n程序调用Py scenedetect，实现类似达芬奇Resolve和剪映的场景剪切探测/智能镜头分割。\n选择视频文件，是单文件操作。选择视频文件夹，是批量操作。",         'select_error': '错误',         'select_error_message': '请选择一个视频文件或文件夹.',         'threshold_error': '错误',         'threshold_error_message': '阈值必须是一个有效的数字.',         'completion_message': '视频分割已完成！分镜视频位于 {} 文件夹内。\n',         'about_info': '感谢使用本程序！\n'     },     'en': {         'title': 'Scene Detection and Splitting 1.3',         'about': 'About',         'select_video_file': 'Select Video File',         'select_video_folder': 'Select Video Folder',         'threshold': 'Scene Change Threshold:',         'start': 'Start Splitting Scenes',         'open_folder': 'Open Output Folder',         'readme': 'September 22, 2024 52pj: rootcup\n Description: The smaller and more sensitive the number of screen change threshold, the more the shot will be cut. The program calls Py scenedetect to implement scene clipping detection/smart shot segmentation similar to Da Vinci Resolve and clipping. \n Select video file, is a single file operation. Select the video folder in batches',         'select_error': 'Error',         'select_error_message': 'Please select a video file or folder.',         'threshold_error': 'Error',         'threshold_error_message': 'Threshold must be a valid number.',         'completion_message': 'Video splitting is complete! The split scenes are located in the {} folder.\n',         'about_info': 'Thank you for using this program!\n'     } }   current_lang = 'zh'  # 当前语言设置     # 加载配置 def load_config():     if os.path.exists(config_file):         with open(config_file, 'r', encoding='utf-8') as file:             config = json.load(file)             return config     return {'threshold': 27, 'last_folder': '', 'language': 'zh'}     # 保存配置 def save_config(config):     with open(config_file, 'w', encoding='utf-8') as file:         json.dump(config, file, ensure_ascii=False, indent=4)     config = load_config() current_lang = config.get('language', 'zh')   # 设置应用程序的默认DPI，确保高DPI屏幕上的显示效果 windll.shcore.SetProcessDpiAwareness(2)  # DPI_AWARENESS_PER_MONITOR_V2 font = 'simhei.ttf' readme = languages[current_lang]['readme']   # 52logo图片的Base64编码 jpg1_base64 = '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' # 创建主窗口 root = tk.Tk() root.title(languages[current_lang]['title']) root.geometry('550x600') root.configure(bg='#f0f0f0')  # 设置背景色   # 创建一个Frame用于放置关于按钮 top_frame = tk.Frame(root, bg='#f0f0f0') top_frame.pack(side=tk.TOP, fill=tk.X)   # 在右上角添加关于按钮 btn_about = tk.Button(top_frame, text=languages[current_lang]['about'], fg='black', font=(font, 10, 'bold'),                       relief=tk.FLAT, command=lambda: show_about_window()) btn_about.pack(side=tk.RIGHT, padx=10, pady=10)   # 创建语言切换按钮 language_menu = tk.Menubutton(top_frame, text="Language", fg='black', font=(font, 10, 'bold'), relief=tk.FLAT) language_menu.menu = tk.Menu(language_menu, tearoff=0) language_menu["menu"] = language_menu.menu language_menu.menu.add_command(label="中文", command=lambda: change_language('zh')) language_menu.menu.add_command(label="English", command=lambda: change_language('en')) language_menu.pack(side=tk.RIGHT, padx=10, pady=10)     # 更新语言 def update_language():     root.title(languages[current_lang]['title'])     btn_about.config(text=languages[current_lang]['about'])     btn_select_file.config(text=languages[current_lang]['select_video_file'])     btn_select_folder.config(text=languages[current_lang]['select_video_folder'])     label_threshold.config(text=languages[current_lang]['threshold'])     btn_start.config(text=languages[current_lang]['start'])     btn_open_folder.config(text=languages[current_lang]['open_folder'])     text_video_paths.delete(1.0, tk.END)     text_video_paths.insert(tk.END, languages[current_lang]['readme'], "tag_1")     # 启动按钮和打开输出文件夹按钮放在同一行的 Frame button0_frame = tk.Frame(root, bg='#f0f0f0') button0_frame.pack(pady=20)   # 选择文件按钮 btn_select_file = tk.Button(button0_frame, text=languages[current_lang]['select_video_file'], bg='#4CAF50', fg='white',                             font=(font, 12, 'bold'),                             relief=tk.FLAT, command=lambda: select_file()) btn_select_file.pack(pady=20, side=tk.LEFT, padx=10)   # 选择文件夹按钮 btn_select_folder = tk.Button(button0_frame, text=languages[current_lang]['select_video_folder'], bg='#4CAF50',                               fg='white', font=(font, 12, 'bold'),                               relief=tk.FLAT, command=lambda: select_folder()) btn_select_folder.pack(pady=10, side=tk.RIGHT, padx=10)   # 设置阈值标签和输入框 label_threshold = tk.Label(root, text=languages[current_lang]['threshold'], bg='#f0f0f0', font=(font, 10)) label_threshold.pack() entry_threshold = tk.Entry(root, width=17) entry_threshold.insert(0, str(config.get('threshold', 27))) entry_threshold.pack(pady=5)   # 启动按钮和打开输出文件夹按钮放在同一行的 Frame button_frame = tk.Frame(root, bg='#f0f0f0') button_frame.pack(pady=20)   # 启动按钮 btn_start = tk.Button(button_frame, text=languages[current_lang]['start'], bg='#2196F3', fg='white',                       font=(font, 12, 'bold'),                       relief=tk.FLAT, command=lambda: start_detection()) btn_start.pack(side=tk.LEFT, padx=10)   # 打开输出文件夹按钮 btn_open_folder = tk.Button(button_frame, text=languages[current_lang]['open_folder'], bg='#FFC107', fg='white',                             font=(font, 12, 'bold'),                             relief=tk.FLAT, command=lambda: open_output_folder()) btn_open_folder.pack(side=tk.LEFT, padx=10)   # 视频路径显示文本框 text_video_paths = tk.Text(root, height=4000, width=6000, bg='#f0f0f0', font=(font, 10)) text_video_paths.pack(padx=5, pady=10) text_video_paths.tag_config("tag_1", foreground="green")  # 设置文本颜色为绿色 text_video_paths.tag_config("tag_2", foreground="blue")   text_video_paths.insert(tk.END, readme, "tag_1")   # 全局变量 selected_folder = config.get('last_folder', '') video_files = []     # 函数定义 def select_file():     global video_files     video_files = [filedialog.askopenfilename(filetypes=[("MP4文件", "*.mp4;*.MP4")])]     if video_files:         update_video_paths()     def select_folder():     global selected_folder, video_files     selected_folder = filedialog.askdirectory()     if selected_folder:         config['last_folder'] = selected_folder         save_config(config)         video_files = [os.path.join(selected_folder, f) for f in os.listdir(selected_folder) if                        f.lower().endswith('.mp4')]         update_video_paths()     def update_video_paths():     text_video_paths.delete(1.0, tk.END)     for video_file in video_files:         text_video_paths.insert(tk.END, video_file + '\n')     def start_detection():     if not video_files:         messagebox.showerror(languages[current_lang]['select_error'], languages[current_lang]['select_error_message'])         return       try:         threshold = float(entry_threshold.get())         config['threshold'] = threshold         save_config(config)     except ValueError:         messagebox.showerror(languages[current_lang]['threshold_error'],                              languages[current_lang]['threshold_error_message'])         return       detection_thread = threading.Thread(target=run_detection, args=(threshold,))     detection_thread.start()     def run_detection(threshold):     with concurrent.futures.ThreadPoolExecutor(max_workers=4) as executor:         futures = [executor.submit(split_video_into_scenes, video_path, threshold) for video_path in video_files]         for future in concurrent.futures.as_completed(futures):             try:                 future.result()             except Exception as e:                 messagebox.showerror(languages[current_lang]['select_error'],                                      f"{languages[current_lang]['select_error_message']} {str(e)}")       text_video_paths.insert(tk.END, '已处理完成全部\n', "tag_1")     def split_video_into_scenes(video_path, threshold=27.0):     base_filename = os.path.splitext(os.path.basename(video_path))[0]     save_path = f'./分镜切割/{base_filename}/'     if not os.path.exists(save_path):         os.makedirs(save_path)       output_file_template = f'{save_path}Scene-$SCENE_NUMBER.mp4'       video = open_video(video_path)     scene_manager = SceneManager()     scene_manager.add_detector(ContentDetector(threshold=threshold))     scene_manager.detect_scenes(video, show_progress=True)     scene_list = scene_manager.get_scene_list()       split_video_ffmpeg(video_path, scene_list, output_file_template=output_file_template, show_progress=True,                        show_output=True)     update_completion_message(save_path)     def update_completion_message(save_path):     message = languages[current_lang]['completion_message'].format(save_path)     text_video_paths.insert(tk.END, message, 'tag_2')     def open_output_folder():     if video_files:         base_filename = os.path.splitext(os.path.basename(video_files[0]))[0]         save_path = f'./分镜切割/{base_filename}/'         if os.path.exists(save_path):             subprocess.Popen(f'explorer "{os.path.abspath(save_path)}"')     def show_about_window():     about_window = Toplevel(root)     about_window.title(languages[current_lang]['about'])     about_window.geometry("300x300")       # 图片和介绍     info_label = tk.Label(about_window, text=languages[current_lang]['about_info'], font=(font, 12))     info_label.pack(pady=10)       # 显示第一张图片     img1_data = base64.b64decode(jpg1_base64)     img1 = Image.open(io.BytesIO(img1_data))     img1_tk = ImageTk.PhotoImage(img1)     img1_label = tk.Label(about_window, image=img1_tk)     img1_label.image = img1_tk     img1_label.pack(pady=5)     # 切换语言 def change_language(lang):     global current_lang     current_lang = lang     config['language'] = current_lang     save_config(config)     update_language()     # 初始化语言 update_language()   # 运行主循环 root.mainloop()

打包好的exe下载:https://wwmd.lanzouv.com/i9clx2ajwunc
密码:52pj

lamquan

使用TransNetV2用CUDA加速，更快更好

q502986650

你好，楼主，首先感谢无私，然后我下载了1.3和1.4，都同样问题：选择了视频，开始分割，显示完成，但是输出文件夹里没有任何文件

wkdxz

很不错，让我们可以免费使用VIP功能

文西思密达

这个 功能很实用！

dappeng

太好了，学习了

W168888

感谢，很有用

SU150228

我有个想法：能否实现自动识别声音停顿切换

abpyu

谢谢大佬分享

rootcup

lamquan 发表于 2024-9-22 20:00
使用TransNetV2用CUDA加速，更快更好

感谢，我去看看

hanbazhen

lamquan 发表于 2024-9-22 20:00
使用TransNetV2用CUDA加速，更快更好

这是一个剪辑软件的名字嘛