{
  "id": 128946,
  "title": "ffmpeg vs cv2 for frames extraction",
  "url": "/competitions/deepfake-detection-challenge/discussion/128946",
  "author_name": "MaximKa",
  "post_date": "2020-02-04T14:05:29.912000",
  "votes": 0,
  "comment_count": 7,
  "views": 0,
  "content": "<p>Are there any advantages to using one instead of the other for frames extraction?</p>",
  "messages": [
    {
      "id": 736959,
      "postDate": "2020-02-04T18:36:33.157Z",
      "content": "<p>I just tried out both of them head to head yesterday and like <a href=\"/dagnelies\">@dagnelies</a>, found that OpenCV was slightly faster for the task of extracting frames from the videos.</p>",
      "rawMarkdown": "I just tried out both of them head to head yesterday and like @dagnelies, found that OpenCV was slightly faster for the task of extracting frames from the videos.",
      "votes": 1
    },
    {
      "id": 736870,
      "postDate": "2020-02-04T16:52:48.663Z",
      "content": "<p>OpenCV seems to have a memory leak (as discussed elsewhere in this competition). This might be an issue if you're planning to use all 16 GB of RAM. 😄 </p>",
      "rawMarkdown": "OpenCV seems to have a memory leak (as discussed elsewhere in this competition). This might be an issue if you're planning to use all 16 GB of RAM. 😄 ",
      "votes": 2
    },
    {
      "id": 995827,
      "postDate": "2020-09-02T20:25:44.657Z",
      "content": "<p>Hi, </p>\n<p>Did anyone find a solution to cv2's memory leak problem? I have a 32 gig RAM which cannot cope up with the linearly increasing memory usage from my face detection script. I have tried explicit garbage collection and also tried to del unused objects. Any leads are highly appreciated. Thanks</p>",
      "rawMarkdown": "Hi, \n\nDid anyone find a solution to cv2's memory leak problem? I have a 32 gig RAM which cannot cope up with the linearly increasing memory usage from my face detection script. I have tried explicit garbage collection and also tried to del unused objects. Any leads are highly appreciated. Thanks"
    },
    {
      "id": 736920,
      "postDate": "2020-02-04T17:56:38.883Z",
      "content": "<p>OpenCV is also included by default on kaggle, unlike ffmpeg which you have to install as extra dependency. ...I think I also tried both out quite some time ago ...if I remember right, OpenCV was actually faster, so I saw no reason to move over to ffmpeg.</p>\n\n<p>However, if you setup a VM or so, ffmpeg might be a bit easier to install and use directly from the command line and shell scripts.</p>",
      "rawMarkdown": "OpenCV is also included by default on kaggle, unlike ffmpeg which you have to install as extra dependency. ...I think I also tried both out quite some time ago ...if I remember right, OpenCV was actually faster, so I saw no reason to move over to ffmpeg.\n\nHowever, if you setup a VM or so, ffmpeg might be a bit easier to install and use directly from the command line and shell scripts."
    },
    {
      "id": 736739,
      "postDate": "2020-02-04T14:12:26.513Z",
      "content": "<p>I find cv2 more simple to use.</p>",
      "rawMarkdown": "I find cv2 more simple to use.",
      "replies": [
        {
          "id": 742692,
          "postDate": "2020-02-11T13:42:38.273Z",
          "content": "<p>How did you get around cv2 memory issue? There seems to be a leakage during frame extraction</p>",
          "rawMarkdown": "How did you get around cv2 memory issue? There seems to be a leakage during frame extraction",
          "votes": 1
        }
      ]
    },
    {
      "id": 736732,
      "postDate": "2020-02-04T14:05:29.913Z",
      "content": "<p>Are there any advantages to using one instead of the other for frames extraction?</p>",
      "rawMarkdown": "Are there any advantages to using one instead of the other for frames extraction?"
    },
    {
      "id": 736921,
      "postDate": "2020-02-04T17:58:42.623Z",
      "rawMarkdown": "",
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 736959,
      "author_name": "Boris Latch",
      "author_url": "",
      "post_date": "2020-02-04T18:36:33.157000",
      "content": "<p>I just tried out both of them head to head yesterday and like <a href=\"/dagnelies\">@dagnelies</a>, found that OpenCV was slightly faster for the task of extracting frames from the videos.</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 736870,
      "author_name": "Human Analog",
      "author_url": "",
      "post_date": "2020-02-04T16:52:48.663000",
      "content": "<p>OpenCV seems to have a memory leak (as discussed elsewhere in this competition). This might be an issue if you're planning to use all 16 GB of RAM. 😄 </p>",
      "votes": 2,
      "replies": []
    },
    {
      "id": 995827,
      "author_name": "Shivangee Trivedi",
      "author_url": "",
      "post_date": "2020-09-02T20:25:44.657000",
      "content": "<p>Hi, </p>\n<p>Did anyone find a solution to cv2's memory leak problem? I have a 32 gig RAM which cannot cope up with the linearly increasing memory usage from my face detection script. I have tried explicit garbage collection and also tried to del unused objects. Any leads are highly appreciated. Thanks</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 736920,
      "author_name": "dagnelies",
      "author_url": "",
      "post_date": "2020-02-04T17:56:38.883000",
      "content": "<p>OpenCV is also included by default on kaggle, unlike ffmpeg which you have to install as extra dependency. ...I think I also tried both out quite some time ago ...if I remember right, OpenCV was actually faster, so I saw no reason to move over to ffmpeg.</p>\n\n<p>However, if you setup a VM or so, ffmpeg might be a bit easier to install and use directly from the command line and shell scripts.</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 736739,
      "author_name": "Ankit Saini",
      "author_url": "",
      "post_date": "2020-02-04T14:12:26.513000",
      "content": "<p>I find cv2 more simple to use.</p>",
      "votes": 0,
      "replies": [
        {
          "id": 742692,
          "author_name": "petkovacs",
          "author_url": "",
          "post_date": "2020-02-11T13:42:38.273000",
          "content": "<p>How did you get around cv2 memory issue? There seems to be a leakage during frame extraction</p>",
          "votes": 1,
          "replies": []
        }
      ]
    },
    {
      "id": 736921,
      "author_name": "",
      "author_url": "",
      "post_date": "2020-02-04T17:58:42.623000",
      "content": "",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "736959": "I just tried out both of them head to head yesterday and like @dagnelies, found that OpenCV was slightly faster for the task of extracting frames from the videos.",
    "736870": "OpenCV seems to have a memory leak (as discussed elsewhere in this competition). This might be an issue if you're planning to use all 16 GB of RAM. 😄 ",
    "995827": "Hi, \n\nDid anyone find a solution to cv2's memory leak problem? I have a 32 gig RAM which cannot cope up with the linearly increasing memory usage from my face detection script. I have tried explicit garbage collection and also tried to del unused objects. Any leads are highly appreciated. Thanks",
    "736920": "OpenCV is also included by default on kaggle, unlike ffmpeg which you have to install as extra dependency. ...I think I also tried both out quite some time ago ...if I remember right, OpenCV was actually faster, so I saw no reason to move over to ffmpeg.\n\nHowever, if you setup a VM or so, ffmpeg might be a bit easier to install and use directly from the command line and shell scripts.",
    "736739": "I find cv2 more simple to use.",
    "736732": "Are there any advantages to using one instead of the other for frames extraction?",
    "736921": ""
  }
}