![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2637251%2Fc1794af7f9a500a3cef7b91681cd9b10%2F.JPG?generation=1579365230542247&alt=media\")
Hi, I have an idea that pre-processing the videos to extract the faces, and store the faces at the path ( /kaggle/working/test_faces ) which I make.
\n\nFinally, using the test_faces folder as input to predict the fake probability.
\n\nSo, I have a question that I use the /kaggle/working as temporary space whether can work?
\n\nThank you!
", "messages": [ { "id": 722466, "postDate": "2020-01-18T15:57:41.480Z", "content": "You can do it but you will not be able to generate a successful submission.
\n\nIf you use the space when you commit and than try to submit - the use of temporary space will max you out on at least 1 of the kaggle limits and you will get a submission error. Not sure what the number is, but a maximum exists for the number of files that can be created and saved in the working space.
", "rawMarkdown": "You can do it but you will not be able to generate a successful submission.\n\nIf you use the space when you commit and than try to submit - the use of temporary space will max you out on at least 1 of the kaggle limits and you will get a submission error. Not sure what the number is, but a maximum exists for the number of files that can be created and saved in the working space.", "votes": 2, "replies": [ { "id": 722485, "postDate": "2020-01-18T16:19:01.270Z", "content": "If I delete the folder before committing, will it be successful??
", "rawMarkdown": "If I delete the folder before committing, will it be successful?? " }, { "id": 722565, "postDate": "2020-01-18T18:13:22.983Z", "content": "If you get to greater than the allowed max the submission will fail - Error message will be like most it's not very descriptive - think I have seen the same error message for all my failures.
\n\nI think the max number of files is around 500 but I have never found any documentation of the exact number.
\n\nSo I think for most models we are limited to doing one video at a time rather than trying a bunch. If you extract 300 frames from a single video and than extract 300 faces there is some chance that saving all those in working will result in you crossing that file count limit.
", "rawMarkdown": "If you get to greater than the allowed max the submission will fail - Error message will be like most it's not very descriptive - think I have seen the same error message for all my failures. \n\nI think the max number of files is around 500 but I have never found any documentation of the exact number.\n\nSo I think for most models we are limited to doing one video at a time rather than trying a bunch. If you extract 300 frames from a single video and than extract 300 faces there is some chance that saving all those in working will result in you crossing that file count limit.", "votes": 2 }, { "id": 723223, "postDate": "2020-01-19T17:07:09.253Z", "content": "I see.\nThank you!
", "rawMarkdown": "I see.\nThank you!" } ] }, { "id": 722341, "postDate": "2020-01-18T12:25:28.497Z", "content": "Hi, I have an idea that pre-processing the videos to extract the faces, and store the faces at the path ( /kaggle/working/test_faces ) which I make.
\n\nFinally, using the test_faces folder as input to predict the fake probability.
\n\nSo, I have a question that I use the /kaggle/working as temporary space whether can work?
\n\nThank you!
", "rawMarkdown": "Hi, I have an idea that pre-processing the videos to extract the faces, and store the faces at the path ( /kaggle/working/test_faces ) which I make.\n\nFinally, using the test_faces folder as input to predict the fake probability.\n\nSo, I have a question that I use the /kaggle/working as temporary space whether can work?\n\nThank you!" }, { "id": 722399, "postDate": "2020-01-18T14:23:37.597Z", "content": "Yes, you can. But after submission.csv file generated, in /kaggle/working directory. You should delete your folders that you add before (like \"test_faces\" ).
\n\nimport shutil\nshutil.rmtree(FOLDER_IN_KAGGLE_WORKING_PATH)
thank you! I will try this.
", "rawMarkdown": "thank you! I will try this." }, { "id": 722510, "postDate": "2020-01-18T16:38:58.797Z", "content": "Hi, Kutsal Baran Özkurt.
\n\nActually, we did this by using /kaggle/working/test_faces to detect the fake probability as the follow.
\n\n
We mapped filename to the submission and filled the label, and we deleted the folder.
\n\nBut we got the score 17.26938.....,we don't know the reasons.
\n\nEven we edited the code as the follow, and got the score 17.26938 too.
\n\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2637251%2Fa2f1961d60acb72876c1deb355d6e404%2F.JPG?generation=1579365773044316&alt=media\")
So we are thinking about whether we used the use the /kaggle/working as temporary space so that we got the score 17.26938.
\n\nWhat do you think? Thanks for your response.
", "rawMarkdown": "Hi, Kutsal Baran Özkurt.\n\nActually, we did this by using /kaggle/working/test_faces to detect the fake probability as the follow.\n\n\n\nWe mapped filename to the submission and filled the label, and we deleted the folder.\n\nBut we got the score 17.26938.....,we don't know the reasons.\n\nEven we edited the code as the follow, and got the score 17.26938 too.\n\n\n\nSo we are thinking about whether we used the use the /kaggle/working as temporary space so that we got the score 17.26938.\n\nWhat do you think? Thanks for your response.\n" }, { "id": 722564, "postDate": "2020-01-18T18:10:33.053Z", "content": "score of 17.26 indicates all your predictions were 1.0 or 0.0 when the script was run against the full test private set.
", "rawMarkdown": "score of 17.26 indicates all your predictions were 1.0 or 0.0 when the script was run against the full test private set.", "votes": 1 }, { "id": 723217, "postDate": "2020-01-19T16:59:50.450Z", "content": "That is strange…\nAll 400 videos had been predicted successfully…\nMy namelstglobal was generated by the input video folder.\nAnd if the system change to the test private dataset, it should work...
", "rawMarkdown": "That is strange…\nAll 400 videos had been predicted successfully…\nMy namelstglobal was generated by the input video folder.\nAnd if the system change to the test private dataset, it should work..." } ] }, { "id": 722378, "postDate": "2020-01-18T13:48:30.900Z", "rawMarkdown": "", "isDeleted": true } ], "comments": [ { "id": 722466, "author_name": "PC Jimmmy", "author_url": "", "post_date": "2020-01-18T15:57:41.480000", "content": "You can do it but you will not be able to generate a successful submission.
\n\nIf you use the space when you commit and than try to submit - the use of temporary space will max you out on at least 1 of the kaggle limits and you will get a submission error. Not sure what the number is, but a maximum exists for the number of files that can be created and saved in the working space.
", "votes": 2, "replies": [ { "id": 722485, "author_name": "ISLAB", "author_url": "", "post_date": "2020-01-18T16:19:01.270000", "content": "If I delete the folder before committing, will it be successful??
", "votes": 0, "replies": [] }, { "id": 722565, "author_name": "PC Jimmmy", "author_url": "", "post_date": "2020-01-18T18:13:22.983000", "content": "If you get to greater than the allowed max the submission will fail - Error message will be like most it's not very descriptive - think I have seen the same error message for all my failures.
\n\nI think the max number of files is around 500 but I have never found any documentation of the exact number.
\n\nSo I think for most models we are limited to doing one video at a time rather than trying a bunch. If you extract 300 frames from a single video and than extract 300 faces there is some chance that saving all those in working will result in you crossing that file count limit.
", "votes": 2, "replies": [] }, { "id": 723223, "author_name": "ISLAB", "author_url": "", "post_date": "2020-01-19T17:07:09.253000", "content": "I see.\nThank you!
", "votes": 0, "replies": [] } ] }, { "id": 722399, "author_name": "", "author_url": "", "post_date": "2020-01-18T14:23:37.597000", "content": "Yes, you can. But after submission.csv file generated, in /kaggle/working directory. You should delete your folders that you add before (like \"test_faces\" ).
\n\nimport shutil\nshutil.rmtree(FOLDER_IN_KAGGLE_WORKING_PATH)
thank you! I will try this.
", "votes": 0, "replies": [] }, { "id": 722510, "author_name": "jerry lo", "author_url": "", "post_date": "2020-01-18T16:38:58.797000", "content": "Hi, Kutsal Baran Özkurt.
\n\nActually, we did this by using /kaggle/working/test_faces to detect the fake probability as the follow.
\n\n
We mapped filename to the submission and filled the label, and we deleted the folder.
\n\nBut we got the score 17.26938.....,we don't know the reasons.
\n\nEven we edited the code as the follow, and got the score 17.26938 too.
\n\n
So we are thinking about whether we used the use the /kaggle/working as temporary space so that we got the score 17.26938.
\n\nWhat do you think? Thanks for your response.
", "votes": 0, "replies": [] }, { "id": 722564, "author_name": "PC Jimmmy", "author_url": "", "post_date": "2020-01-18T18:10:33.053000", "content": "score of 17.26 indicates all your predictions were 1.0 or 0.0 when the script was run against the full test private set.
", "votes": 1, "replies": [] }, { "id": 723217, "author_name": "jerry lo", "author_url": "", "post_date": "2020-01-19T16:59:50.450000", "content": "That is strange…\nAll 400 videos had been predicted successfully…\nMy namelstglobal was generated by the input video folder.\nAnd if the system change to the test private dataset, it should work...
", "votes": 0, "replies": [] } ] }, { "id": 722378, "author_name": "", "author_url": "", "post_date": "2020-01-18T13:48:30.900000", "content": "", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "722466": "You can do it but you will not be able to generate a successful submission.\n\nIf you use the space when you commit and than try to submit - the use of temporary space will max you out on at least 1 of the kaggle limits and you will get a submission error. Not sure what the number is, but a maximum exists for the number of files that can be created and saved in the working space.", "722341": "Hi, I have an idea that pre-processing the videos to extract the faces, and store the faces at the path ( /kaggle/working/test_faces ) which I make.\n\nFinally, using the test_faces folder as input to predict the fake probability.\n\nSo, I have a question that I use the /kaggle/working as temporary space whether can work?\n\nThank you!", "722399": "Yes, you can. But after submission.csv file generated, in /kaggle/working directory. You should delete your folders that you add before (like \"test_faces\" ). \n\n`import shutil`\n`shutil.rmtree(FOLDER_IN_KAGGLE_WORKING_PATH)`", "722378": "" } }