{
  "id": 17770,
  "title": "Train face detector with Fast R-CNN",
  "url": "/competitions/noaa-right-whale-recognition/discussion/17770",
  "author_name": "",
  "post_date": "2015-12-08T05:43:45.663Z",
  "votes": 6,
  "comment_count": 2,
  "views": 2636,
  "content": "<p>Hi guys,</p>\n\n<p>I wrote something <a href=\"https://github.com/coldmanck/fast-rcnn/blob/master/README.md\">here</a> when I trained a whale face detector with <a href=\"https://github.com/rbgirshick/fast-rcnn\">Fast R-CNN</a> (by Ross Girshick), all of you are welcome to give it a try. I shared it publicly because I've also shared this with other teams. I will appreciate if you could give my repo a star if you find this helpful :) </p>\n\n<p>good luck!</p>",
  "messages": [
    {
      "id": "100527",
      "postDate": "12/08/2015 05:43:45",
      "content": "<p>Hi guys,</p>\n\n<p>I wrote something <a href=\"https://github.com/coldmanck/fast-rcnn/blob/master/README.md\">here</a> when I trained a whale face detector with <a href=\"https://github.com/rbgirshick/fast-rcnn\">Fast R-CNN</a> (by Ross Girshick), all of you are welcome to give it a try. I shared it publicly because I've also shared this with other teams. I will appreciate if you could give my repo a star if you find this helpful :) </p>\n\n<p>good luck!</p>",
      "rawMarkdown": "Hi guys,\r\n\r\nI wrote something [here][1] when I trained a whale face detector with [Fast R-CNN][2] (by Ross Girshick), all of you are welcome to give it a try. I shared it publicly because I've also shared this with other teams. I will appreciate if you could give my repo a star if you find this helpful :) \r\n\r\ngood luck!\r\n\r\n\r\n  [1]: https://github.com/coldmanck/fast-rcnn/blob/master/README.md\r\n  [2]: https://github.com/rbgirshick/fast-rcnn",
      "votes": null
    },
    {
      "id": "100616",
      "postDate": "12/08/2015 23:57:52",
      "content": "<p>Hi @Meng-Jiun Chiou,</p>\n\n<p>Thanks  a lot for sharing.  Does the repo include the results of the run. \nI see lots of bounding boxes in the test.mat  files (in data/kaggle). Are those the boxes predicted by the network?</p>",
      "rawMarkdown": "Hi @Meng-Jiun Chiou,\r\n\r\nThanks  a lot for sharing.  Does the repo include the results of the run. \r\nI see lots of bounding boxes in the test.mat  files (in data/kaggle). Are those the boxes predicted by the network?",
      "votes": null
    },
    {
      "id": "100619",
      "postDate": "12/09/2015 01:56:30",
      "content": "<p>Hi @Vinh Nguyen,</p>\n\n<p>well, they're <strong>box proposals</strong> which are generated by selective search. Fast R-CNN need input of proposals to work. You can refer to the paper <a href=\"http://arxiv.org/abs/1506.01497\">http://arxiv.org/abs/1506.01497</a></p>",
      "rawMarkdown": "Hi @Vinh Nguyen,\r\n\r\nwell, they're **box proposals** which are generated by selective search. Fast R-CNN need input of proposals to work. You can refer to the paper http://arxiv.org/abs/1506.01497",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 100616,
      "author_name": "vinhnguyen",
      "author_url": "",
      "post_date": "12/08/2015 23:57:52",
      "content": "<p>Hi @Meng-Jiun Chiou,</p>\n\n<p>Thanks  a lot for sharing.  Does the repo include the results of the run. \nI see lots of bounding boxes in the test.mat  files (in data/kaggle). Are those the boxes predicted by the network?</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 100619,
      "author_name": "coldmanck",
      "author_url": "",
      "post_date": "12/09/2015 01:56:30",
      "content": "<p>Hi @Vinh Nguyen,</p>\n\n<p>well, they're <strong>box proposals</strong> which are generated by selective search. Fast R-CNN need input of proposals to work. You can refer to the paper <a href=\"http://arxiv.org/abs/1506.01497\">http://arxiv.org/abs/1506.01497</a></p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "100527": "Hi guys,\r\n\r\nI wrote something [here][1] when I trained a whale face detector with [Fast R-CNN][2] (by Ross Girshick), all of you are welcome to give it a try. I shared it publicly because I've also shared this with other teams. I will appreciate if you could give my repo a star if you find this helpful :) \r\n\r\ngood luck!\r\n\r\n\r\n  [1]: https://github.com/coldmanck/fast-rcnn/blob/master/README.md\r\n  [2]: https://github.com/rbgirshick/fast-rcnn",
    "100616": "Hi @Meng-Jiun Chiou,\r\n\r\nThanks  a lot for sharing.  Does the repo include the results of the run. \r\nI see lots of bounding boxes in the test.mat  files (in data/kaggle). Are those the boxes predicted by the network?",
    "100619": "Hi @Vinh Nguyen,\r\n\r\nwell, they're **box proposals** which are generated by selective search. Fast R-CNN need input of proposals to work. You can refer to the paper http://arxiv.org/abs/1506.01497"
  },
  "source": "meta"
}