使用 FFmpeg 和 Python，根据视频内容来找出重复的视频

视频去重.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "40f15af6",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-10-19T06:56:22.127623Z",
     "start_time": "2021-10-19T06:56:22.086793Z"
    }
   },
   "outputs": [],
   "source": [
    "import ffmpeg\n",
    "from diffimg import diff\n",
    "from glob import glob\n",
    "from IPython.display import Image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "46c97367",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-10-19T06:56:22.131302Z",
     "start_time": "2021-10-19T06:56:22.129020Z"
    }
   },
   "outputs": [],
   "source": [
    "video_path = '/home/tee/temp/工作/视频2/*.mp4'\n",
    "img_path = '/home/tee/temp/工作/去重截图/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "d522b039",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-10-19T06:56:22.135920Z",
     "start_time": "2021-10-19T06:56:22.132984Z"
    }
   },
   "outputs": [],
   "source": [
    "mp4s = glob(video_path)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0febcac9",
   "metadata": {},
   "source": [
    "# 文件名去重检查"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "153e85d1",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-10-19T06:56:22.141599Z",
     "start_time": "2021-10-19T06:56:22.137799Z"
    }
   },
   "outputs": [],
   "source": [
    "mp4s_name = [mp4.split('/')[-1] for mp4 in mp4s]\n",
    "\n",
    "assert set([x for x in mp4s_name if mp4s_name.count(x) > 1]) == set()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "205e3ff6",
   "metadata": {},
   "source": [
    "# 取样截图"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "ba3cf54a",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-10-19T06:56:22.146402Z",
     "start_time": "2021-10-19T06:56:22.142494Z"
    }
   },
   "outputs": [],
   "source": [
    "def save_img(q, q_name):\n",
    "    out, _ = (\n",
    "        ffmpeg\n",
    "        .input(mp4, ss=q)\n",
    "        .filter('select', 'gte(n,{})'.format(1))\n",
    "        .output('pipe:', vframes=1, format='image2', vcodec='mjpeg')\n",
    "        .global_args('-loglevel', 'error')\n",
    "        .global_args('-y')\n",
    "        .run(capture_stdout=True)\n",
    "    )\n",
    "\n",
    "    with open(img_path + mp4.split('/')[-1][:-4] + f'_{q_name}.jpg', 'wb') as f:\n",
    "        f.write(out)\n",
    "    return out"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "27166435",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-10-19T06:56:28.653726Z",
     "start_time": "2021-10-19T06:56:22.147624Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[处理中] ... /home/tee/temp/工作/视频2/横版_5.mp4\n",
      "[处理中] ... /home/tee/temp/工作/视频2/横版_1.mp4\n",
      "[处理中] ... /home/tee/temp/工作/视频2/横版_2.mp4\n",
      "[处理中] ... /home/tee/temp/工作/视频2/横版_3.mp4\n",
      "[处理中] ... /home/tee/temp/工作/视频2/横版_7.mp4\n",
      "[处理中] ... /home/tee/temp/工作/视频2/横版_4.mp4\n",
      "[处理中] ... /home/tee/temp/工作/视频2/横版_6.mp4\n"
     ]
    }
   ],
   "source": [
    "for mp4 in mp4s:\n",
    "    \n",
    "    print(f'[处理中] ... {mp4}')\n",
    "    \n",
    "    duration = float(ffmpeg.probe(mp4)['streams'][0]['duration'])\n",
    "    q_add_on = duration / 5\n",
    "\n",
    "    q1 = q_add_on\n",
    "    q2 = q1 + q_add_on\n",
    "    q3 = q2 + q_add_on\n",
    "    q4 = q3 + q_add_on\n",
    "\n",
    "    save_img(q1, 'q1')\n",
    "    save_img(q2, 'q2')\n",
    "    save_img(q3, 'q3')\n",
    "    save_img(q4, 'q4')\n",
    "    \n",
    "#     break"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "1c5ce132",
   "metadata": {},
   "source": [
    "# diff 图片"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "e36aaac8",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-10-19T06:56:28.658794Z",
     "start_time": "2021-10-19T06:56:28.655072Z"
    }
   },
   "outputs": [],
   "source": [
    "q1_list = glob(f'{img_path}*_q1.jpg')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "f7d2bca2",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-10-19T06:56:28.671874Z",
     "start_time": "2021-10-19T06:56:28.661737Z"
    }
   },
   "outputs": [],
   "source": [
    "def diff_list(q1_list, ori_str, dst_str):\n",
    "    q2_list = []\n",
    "    for q1 in q1_list[1:]:\n",
    "        print(f'[检查 {ori_str} 中] ... {q1}')\n",
    "        if diff(q1_list[0], q1) < 0.1:\n",
    "            q2_list.append(q1)\n",
    "    q2_list = [q1_list[0]] + q2_list\n",
    "    q2_list = [q2.replace(ori_str, dst_str) for q2 in q2_list]\n",
    "    \n",
    "    return q2_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "bf615c4b",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-10-19T06:56:30.071298Z",
     "start_time": "2021-10-19T06:56:28.673494Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[检查 _q1.jpg 中] ... /home/tee/temp/工作/去重截图/横版_5_q1.jpg\n",
      "[检查 _q1.jpg 中] ... /home/tee/temp/工作/去重截图/横版_4_q1.jpg\n",
      "[检查 _q1.jpg 中] ... /home/tee/temp/工作/去重截图/横版_7_q1.jpg\n",
      "[检查 _q1.jpg 中] ... /home/tee/temp/工作/去重截图/横版_3_q1.jpg\n",
      "[检查 _q1.jpg 中] ... /home/tee/temp/工作/去重截图/横版_1_q1.jpg\n",
      "[检查 _q1.jpg 中] ... /home/tee/temp/工作/去重截图/横版_6_q1.jpg\n"
     ]
    }
   ],
   "source": [
    "q2_list = diff_list(q1_list, '_q1.jpg', '_q2.jpg')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "bb3f51eb",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-10-19T06:56:30.153994Z",
     "start_time": "2021-10-19T06:56:30.073801Z"
    },
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[检查 _q2.jpg 中] ... /home/tee/temp/工作/去重截图/横版_7_q2.jpg\n",
      "[检查 _q2.jpg 中] ... /home/tee/temp/工作/去重截图/横版_6_q2.jpg\n"
     ]
    }
   ],
   "source": [
    "q3_list = diff_list(q2_list, '_q2.jpg', '_q3.jpg')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "78121dec",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-10-19T06:56:30.248378Z",
     "start_time": "2021-10-19T06:56:30.155230Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[检查 _q3.jpg 中] ... /home/tee/temp/工作/去重截图/横版_7_q3.jpg\n",
      "[检查 _q3.jpg 中] ... /home/tee/temp/工作/去重截图/横版_6_q3.jpg\n"
     ]
    }
   ],
   "source": [
    "q4_list = diff_list(q3_list, '_q3.jpg', '_q4.jpg')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "3a2698ce",
   "metadata": {},
   "source": [
    "# diff 结果"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "89d50aee",
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-10-19T06:56:30.253653Z",
     "start_time": "2021-10-19T06:56:30.249814Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "有重复视频\n",
      "\n",
      "['横版_2', '横版_7', '横版_6']\n"
     ]
    }
   ],
   "source": [
    "if q4_list == []:\n",
    "    print('无重复视频')\n",
    "else:\n",
    "    duplicate_video = [v.split('/')[-1][:-7] for v in q4_list]\n",
    "    print('有重复视频')\n",
    "    print()\n",
    "    print(duplicate_video)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e9542932",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}