{
  "id": 216542,
  "title": "How to submit the csv prediction file?",
  "url": "/competitions/cassava-leaf-disease-classification/discussion/216542",
  "author_name": "",
  "post_date": "2021-02-03T06:41:03.687109300Z",
  "votes": 2,
  "comment_count": 9,
  "views": 0,
  "content": "<p>I am new to Kaggle, sorry I am really confused about this question.<br>\nIn this competition, I have already got one H5 file(trained by myself), but there is only one image in the test image folder, how can I submit my prediction result with only one image's predication label? </p>\n<p>Please help me, thanks in advance!</p>",
  "messages": [
    {
      "id": "1183689",
      "postDate": "02/03/2021 06:41:03",
      "content": "<p>I am new to Kaggle, sorry I am really confused about this question.<br>\nIn this competition, I have already got one H5 file(trained by myself), but there is only one image in the test image folder, how can I submit my prediction result with only one image's predication label? </p>\n<p>Please help me, thanks in advance!</p>",
      "rawMarkdown": "I am new to Kaggle, sorry I am really confused about this question.\nIn this competition, I have already got one H5 file(trained by myself), but there is only one image in the test image folder, how can I submit my prediction result with only one image's predication label? \n\nPlease help me, thanks in advance!",
      "votes": null
    },
    {
      "id": "1183690",
      "postDate": "02/03/2021 06:42:11",
      "content": "<p>I used tensorflow2.0 in this competition</p>",
      "rawMarkdown": "I used tensorflow2.0 in this competition",
      "votes": null
    },
    {
      "id": "1183718",
      "postDate": "02/03/2021 07:06:19",
      "content": "<p>Hello there, I'm kinda new too. <br>\nThere are many references out there under \"Code\". <br>\nHowever, you can have a look at my notebook <a href=\"https://www.kaggle.com/nelsonwongisme/tf-keras-dataaugmentation-tensorboard\" target=\"_blank\">Click this</a></p>\n<p>Scroll till the bottom, check the LAST code block. Replace your model path(after upload) into <br>\n<code>model = tf.keras.models.load_model(\"./&lt;HERE HERE HERE&gt;.h5\")</code> then click save and commit.</p>\n<p>After committing it, to submit:<br>\nCompete &gt; Select this Comp &gt; Submit Predictions.</p>\n<p>Do disable your internet connection before \"Save &amp; Commit\".</p>\n<p>Hope this helps!</p>",
      "rawMarkdown": "Hello there, I'm kinda new too. \nThere are many references out there under \"Code\". \nHowever, you can have a look at my notebook [Click this](https://www.kaggle.com/nelsonwongisme/tf-keras-dataaugmentation-tensorboard)\n\nScroll till the bottom, check the LAST code block. Replace your model path(after upload) into \n`model = tf.keras.models.load_model(\"./<HERE HERE HERE>.h5\")` then click save and commit.\n\nAfter committing it, to submit:\nCompete > Select this Comp > Submit Predictions.\n\nDo disable your internet connection before \"Save & Commit\".\n\nHope this helps!",
      "votes": null
    },
    {
      "id": "1183723",
      "postDate": "02/03/2021 07:14:35",
      "content": "<p><a href=\"https://www.kaggle.com/nelsonwongisme\" target=\"_blank\">@nelsonwongisme</a> this helps a lot, thanks!</p>",
      "rawMarkdown": "nelsonwongisme this helps a lot, thanks!",
      "votes": null
    },
    {
      "id": "1183725",
      "postDate": "02/03/2021 07:17:09",
      "content": "<p>Does the order of committing is: run all code---&gt;save  and commit----&gt;compete-----&gt;submit prediction?</p>",
      "rawMarkdown": "Does the order of committing is: run all code--->save  and commit---->compete----->submit prediction?",
      "votes": null
    },
    {
      "id": "1183728",
      "postDate": "02/03/2021 07:20:26",
      "content": "<p>I usually run all codes first to make sure the codes working. As after \"Save &amp; Commit\", it basically runs the codes from top to bottom.</p>\n<p>Haha ya, I should explain in point form though.</p>\n<p>It won't take long tbh(depends on your trained model size), I suggest you just test it out.</p>",
      "rawMarkdown": "I usually run all codes first to make sure the codes working. As after \"Save & Commit\", it basically runs the codes from top to bottom.\n\nHaha ya, I should explain in point form though.\n\nIt won't take long tbh(depends on your trained model size), I suggest you just test it out.",
      "votes": null
    },
    {
      "id": "1183736",
      "postDate": "02/03/2021 07:28:10",
      "content": "<p>Also, to answer your </p>\n<blockquote>\n  <p>but there is only one image in the test image folder, how can I submit my prediction result with only one image's predication label?</p>\n</blockquote>\n<p>Do note that, in <a href=\"https://www.kaggle.com/c/cassava-leaf-disease-classification/data\" target=\"_blank\">here</a></p>\n<p>They mentioned:</p>\n<blockquote>\n  <p>[train/test]_images the image files. The full set of test images will only be available to your notebook when it is submitted for scoring. Expect to see roughly 15,000 images in the test set.</p>\n</blockquote>",
      "rawMarkdown": "Also, to answer your \n> but there is only one image in the test image folder, how can I submit my prediction result with only one image's predication label?\n\nDo note that, in [here](https://www.kaggle.com/c/cassava-leaf-disease-classification/data)\n\nThey mentioned:\n>  [train/test]_images the image files. The full set of test images will only be available to your notebook when it is submitted for scoring. Expect to see roughly 15,000 images in the test set.",
      "votes": null
    },
    {
      "id": "1183758",
      "postDate": "02/03/2021 07:48:34",
      "content": "<p>Thanks!!!!! Can I use my h5 file(already trained) to do the prediction（upload it to the location in kaggle dataset）, and do not write the part of Neural Networks, just load the h5 model to do the prediction? because I have trained my model for about 7 hours, if I click \"save and commit\" to ran it again, it will cost me much time.</p>",
      "rawMarkdown": "Thanks!!!!! Can I use my h5 file(already trained) to do the prediction（upload it to the location in kaggle dataset）, and do not write the part of Neural Networks, just load the h5 model to do the prediction? because I have trained my model for about 7 hours, if I click \"save and commit\" to ran it again, it will cost me much time.",
      "votes": null
    },
    {
      "id": "1183775",
      "postDate": "02/03/2021 07:58:17",
      "content": "<p>Yes yes. Just upload it as dataset, then replace below line with path to that model.<br>\n<code>model = tf.keras.models.load_model(\"./&lt;HERE HERE HERE&gt;.h5\")</code></p>\n<p>Your model path should look like this:<br>\n<code>\"../input/cas1mode/cassava_Model.h5\"</code> </p>\n<p>It will only cost time to upload your model + load_model. </p>",
      "rawMarkdown": "Yes yes. Just upload it as dataset, then replace below line with path to that model.\n`model = tf.keras.models.load_model(\"./<HERE HERE HERE>.h5\")`\n\nYour model path should look like this:\n`\"../input/cas1mode/cassava_Model.h5\"` \n\nIt will only cost time to upload your model + load_model.",
      "votes": null
    },
    {
      "id": "1184357",
      "postDate": "02/03/2021 13:59:00",
      "content": "<p>Created a  post for the same purpose <a href=\"https://www.kaggle.com/c/cassava-leaf-disease-classification/discussion/216461\" target=\"_blank\">here</a></p>",
      "rawMarkdown": "Created a  post for the same purpose [here](https://www.kaggle.com/c/cassava-leaf-disease-classification/discussion/216461)",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1183690,
      "author_name": "mikesong",
      "author_url": "",
      "post_date": "02/03/2021 06:42:11",
      "content": "<p>I used tensorflow2.0 in this competition</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1183718,
      "author_name": "nelsonwongisme",
      "author_url": "",
      "post_date": "02/03/2021 07:06:19",
      "content": "<p>Hello there, I'm kinda new too. <br>\nThere are many references out there under \"Code\". <br>\nHowever, you can have a look at my notebook <a href=\"https://www.kaggle.com/nelsonwongisme/tf-keras-dataaugmentation-tensorboard\" target=\"_blank\">Click this</a></p>\n<p>Scroll till the bottom, check the LAST code block. Replace your model path(after upload) into <br>\n<code>model = tf.keras.models.load_model(\"./&lt;HERE HERE HERE&gt;.h5\")</code> then click save and commit.</p>\n<p>After committing it, to submit:<br>\nCompete &gt; Select this Comp &gt; Submit Predictions.</p>\n<p>Do disable your internet connection before \"Save &amp; Commit\".</p>\n<p>Hope this helps!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1183723,
      "author_name": "mikesong",
      "author_url": "",
      "post_date": "02/03/2021 07:14:35",
      "content": "<p><a href=\"https://www.kaggle.com/nelsonwongisme\" target=\"_blank\">@nelsonwongisme</a> this helps a lot, thanks!</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1183725,
      "author_name": "mikesong",
      "author_url": "",
      "post_date": "02/03/2021 07:17:09",
      "content": "<p>Does the order of committing is: run all code---&gt;save  and commit----&gt;compete-----&gt;submit prediction?</p>",
      "votes": null,
      "replies": [
        {
          "id": 1183728,
          "author_name": "nelsonwongisme",
          "author_url": "",
          "post_date": "02/03/2021 07:20:26",
          "content": "<p>I usually run all codes first to make sure the codes working. As after \"Save &amp; Commit\", it basically runs the codes from top to bottom.</p>\n<p>Haha ya, I should explain in point form though.</p>\n<p>It won't take long tbh(depends on your trained model size), I suggest you just test it out.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1183736,
          "author_name": "nelsonwongisme",
          "author_url": "",
          "post_date": "02/03/2021 07:28:10",
          "content": "<p>Also, to answer your </p>\n<blockquote>\n  <p>but there is only one image in the test image folder, how can I submit my prediction result with only one image's predication label?</p>\n</blockquote>\n<p>Do note that, in <a href=\"https://www.kaggle.com/c/cassava-leaf-disease-classification/data\" target=\"_blank\">here</a></p>\n<p>They mentioned:</p>\n<blockquote>\n  <p>[train/test]_images the image files. The full set of test images will only be available to your notebook when it is submitted for scoring. Expect to see roughly 15,000 images in the test set.</p>\n</blockquote>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1183758,
          "author_name": "mikesong",
          "author_url": "",
          "post_date": "02/03/2021 07:48:34",
          "content": "<p>Thanks!!!!! Can I use my h5 file(already trained) to do the prediction（upload it to the location in kaggle dataset）, and do not write the part of Neural Networks, just load the h5 model to do the prediction? because I have trained my model for about 7 hours, if I click \"save and commit\" to ran it again, it will cost me much time.</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1183775,
          "author_name": "nelsonwongisme",
          "author_url": "",
          "post_date": "02/03/2021 07:58:17",
          "content": "<p>Yes yes. Just upload it as dataset, then replace below line with path to that model.<br>\n<code>model = tf.keras.models.load_model(\"./&lt;HERE HERE HERE&gt;.h5\")</code></p>\n<p>Your model path should look like this:<br>\n<code>\"../input/cas1mode/cassava_Model.h5\"</code> </p>\n<p>It will only cost time to upload your model + load_model. </p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1184357,
      "author_name": "vickygoyal",
      "author_url": "",
      "post_date": "02/03/2021 13:59:00",
      "content": "<p>Created a  post for the same purpose <a href=\"https://www.kaggle.com/c/cassava-leaf-disease-classification/discussion/216461\" target=\"_blank\">here</a></p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1183689": "I am new to Kaggle, sorry I am really confused about this question.\nIn this competition, I have already got one H5 file(trained by myself), but there is only one image in the test image folder, how can I submit my prediction result with only one image's predication label? \n\nPlease help me, thanks in advance!",
    "1183690": "I used tensorflow2.0 in this competition",
    "1183718": "Hello there, I'm kinda new too. \nThere are many references out there under \"Code\". \nHowever, you can have a look at my notebook [Click this](https://www.kaggle.com/nelsonwongisme/tf-keras-dataaugmentation-tensorboard)\n\nScroll till the bottom, check the LAST code block. Replace your model path(after upload) into \n`model = tf.keras.models.load_model(\"./<HERE HERE HERE>.h5\")` then click save and commit.\n\nAfter committing it, to submit:\nCompete > Select this Comp > Submit Predictions.\n\nDo disable your internet connection before \"Save & Commit\".\n\nHope this helps!",
    "1183723": "nelsonwongisme this helps a lot, thanks!",
    "1183725": "Does the order of committing is: run all code--->save  and commit---->compete----->submit prediction?",
    "1183728": "I usually run all codes first to make sure the codes working. As after \"Save & Commit\", it basically runs the codes from top to bottom.\n\nHaha ya, I should explain in point form though.\n\nIt won't take long tbh(depends on your trained model size), I suggest you just test it out.",
    "1183736": "Also, to answer your \n> but there is only one image in the test image folder, how can I submit my prediction result with only one image's predication label?\n\nDo note that, in [here](https://www.kaggle.com/c/cassava-leaf-disease-classification/data)\n\nThey mentioned:\n>  [train/test]_images the image files. The full set of test images will only be available to your notebook when it is submitted for scoring. Expect to see roughly 15,000 images in the test set.",
    "1183758": "Thanks!!!!! Can I use my h5 file(already trained) to do the prediction（upload it to the location in kaggle dataset）, and do not write the part of Neural Networks, just load the h5 model to do the prediction? because I have trained my model for about 7 hours, if I click \"save and commit\" to ran it again, it will cost me much time.",
    "1183775": "Yes yes. Just upload it as dataset, then replace below line with path to that model.\n`model = tf.keras.models.load_model(\"./<HERE HERE HERE>.h5\")`\n\nYour model path should look like this:\n`\"../input/cas1mode/cassava_Model.h5\"` \n\nIt will only cost time to upload your model + load_model.",
    "1184357": "Created a  post for the same purpose [here](https://www.kaggle.com/c/cassava-leaf-disease-classification/discussion/216461)"
  },
  "source": "meta"
}