I have played with dogs classification. Here some practical results - https://medium.com/@kirillpanarin/dog-breed-classification-hands-on-approach-b5e4f88c333e
", "messages": [ { "id": "235538", "postDate": "10/25/2017 17:17:28", "content": "I have played with dogs classification. Here some practical results - https://medium.com/@kirillpanarin/dog-breed-classification-hands-on-approach-b5e4f88c333e
", "rawMarkdown": "I have played with dogs classification. Here some practical results - https://medium.com/@kirillpanarin/dog-breed-classification-hands-on-approach-b5e4f88c333e", "votes": null }, { "id": "235924", "postDate": "10/26/2017 10:19:10", "content": "Thanks for the code, Kirill. I am trying to read it through and on this stage python -m src.data_preparation.stanford_ds_to_tfrecords it looks like one_hot_encoder, _ = dataset.one_hot_label_encoder() expects labels already to be prepared and saved in paths.LABELS, but I cant find a piece of code which should do that.
Hi SM! Thanks for letting me know about the issue. The code expected labels.csv from this competition training dataset. I didn't check it in. I moved the list of all breeds to the new file 'data/breeds.csv' and modified all code to use it instead of loading 'data/train/labels.csv' and taking unique values of breeds. Please, pull the latest changes from the repo and give it a try again - it has to work now.
", "rawMarkdown": "Hi SM! Thanks for letting me know about the issue. The code expected labels.csv from this competition training dataset. I didn't check it in. I moved the list of all breeds to the new file 'data/breeds.csv' and modified all code to use it instead of loading 'data/train/labels.csv' and taking unique values of breeds. Please, pull the latest changes from the repo and give it a try again - it has to work now.", "votes": null }, { "id": "237477", "postDate": "10/30/2017 12:03:50", "content": "", "rawMarkdown": "", "votes": null }, { "id": "237484", "postDate": "10/30/2017 12:29:26", "content": "Thanks for good writing, Kirill. \nI have a question though.\nWhen I execute\n‘python -m src.data_preparation.stanford_ds_to_tfrecords’ in the root of your repo.\nI got a warning\nFile “C:\\Users\\ME\\dog-breeds-classification\\src\\data_preparation\\stanford_ds_to_tfrecords.py”, line 7, in import data\nModule Not Found Error: No module named ‘dataset'
\n\nHow can I fix it? Thanks.
", "rawMarkdown": "Thanks for good writing, Kirill. \nI have a question though.\nWhen I execute\n‘python -m src.data_preparation.stanford_ds_to_tfrecords’ in the root of your repo.\nI got a warning\nFile “C:\\Users\\ME\\dog-breeds-classification\\src\\data_preparation\\stanford_ds_to_tfrecords.py”, line 7, in", "votes": null }, { "id": "237537", "postDate": "10/30/2017 14:41:52", "content": "I clear it.
", "rawMarkdown": "I clear it.", "votes": null }, { "id": "246079", "postDate": "11/20/2017 14:27:25", "content": "Hi folks,
\n\nI am facing a strange error where .py files are getting executed until the tensorflow throws the FileNotFoundError : tensorflow.python.framework.errors_impl.NotFoundError: NewRandomAccessFile failed to Create/Open: frozen/inception/classify_image_graph_def.pb : The system cannot find the path specified`
My guess is : setup is missing from the path since I'm calling out from root directory. Is this correct?
Any idea on how to resolve this?
", "rawMarkdown": "Hi folks,\n\nI am facing a strange error where `.py` files are getting executed until the tensorflow throws the `FileNotFoundError` : `tensorflow.python.framework.errors_impl.NotFoundError: NewRandomAccessFile failed to Create/Open: frozen/inception/classify_image_graph_def.pb` : The system cannot find the path specified` \n\nMy guess is : `setup` is missing from the `path` since I'm calling out from root directory. Is this correct?\n\n\nAny idea on how to resolve this?", "votes": null }, { "id": "246128", "postDate": "11/20/2017 15:56:04", "content": "Hi SM, could you help me understand the tf_record_utils in the stanford_ds_to_tfrecords.py code. I get a module not found error for this one. When trying to install this module using pip I get No matching distribution found for tf_record_utils error
Its a local py file which you can find here dog-breeds-classification/src/data_preparation/tf_record_utils.py
", "rawMarkdown": "Its a local py file which you can find here dog-breeds-classification/src/data_preparation/tf_record_utils.py", "votes": null }, { "id": "246173", "postDate": "11/20/2017 16:21:00", "content": "got it.. So it should be found automatically right when I run the stanford_ds_to_tfrecords.py file. But somehow, its not able to find this.
In which directory am i suppose to run the conversion file? inside the src folder ? I guess, I need to add some folders to the PATH variables
Hi Joel. \nIn 'stanford_ds_to_tfrecords.py' change
\n(line number: 12) from tf_record_utils import * -> from src.data_preparation import tf_record_utils
\nI hope it helps.
Hi DonghunPark, thank you for helping out. It worked for me!\nCould you look into the error i'm facing now(raised it in a separate comment?)
", "rawMarkdown": "Hi DonghunPark, thank you for helping out. It worked for me!\nCould you look into the error i'm facing now(raised it in a separate comment?)", "votes": null } ], "comments": [ { "id": 235924, "author_name": "sermakarevich", "author_url": "", "post_date": "10/26/2017 10:19:10", "content": "Thanks for the code, Kirill. I am trying to read it through and on this stage python -m src.data_preparation.stanford_ds_to_tfrecords it looks like one_hot_encoder, _ = dataset.one_hot_label_encoder() expects labels already to be prepared and saved in paths.LABELS, but I cant find a piece of code which should do that.
Hi SM, could you help me understand the tf_record_utils in the stanford_ds_to_tfrecords.py code. I get a module not found error for this one. When trying to install this module using pip I get No matching distribution found for tf_record_utils error
Its a local py file which you can find here dog-breeds-classification/src/data_preparation/tf_record_utils.py
", "votes": null, "replies": [] }, { "id": 246173, "author_name": "vasaibaby", "author_url": "", "post_date": "11/20/2017 16:21:00", "content": "got it.. So it should be found automatically right when I run the stanford_ds_to_tfrecords.py file. But somehow, its not able to find this.
In which directory am i suppose to run the conversion file? inside the src folder ? I guess, I need to add some folders to the PATH variables
Hi Joel. \nIn 'stanford_ds_to_tfrecords.py' change
\n(line number: 12) from tf_record_utils import * -> from src.data_preparation import tf_record_utils
\nI hope it helps.
Hi DonghunPark, thank you for helping out. It worked for me!\nCould you look into the error i'm facing now(raised it in a separate comment?)
", "votes": null, "replies": [] } ] }, { "id": 236025, "author_name": "kirillpanarin", "author_url": "", "post_date": "10/26/2017 14:40:16", "content": "Hi SM! Thanks for letting me know about the issue. The code expected labels.csv from this competition training dataset. I didn't check it in. I moved the list of all breeds to the new file 'data/breeds.csv' and modified all code to use it instead of loading 'data/train/labels.csv' and taking unique values of breeds. Please, pull the latest changes from the repo and give it a try again - it has to work now.
", "votes": null, "replies": [] }, { "id": 237477, "author_name": "donghp", "author_url": "", "post_date": "10/30/2017 12:03:50", "content": "", "votes": null, "replies": [] }, { "id": 237484, "author_name": "donghp", "author_url": "", "post_date": "10/30/2017 12:29:26", "content": "Thanks for good writing, Kirill. \nI have a question though.\nWhen I execute\n‘python -m src.data_preparation.stanford_ds_to_tfrecords’ in the root of your repo.\nI got a warning\nFile “C:\\Users\\ME\\dog-breeds-classification\\src\\data_preparation\\stanford_ds_to_tfrecords.py”, line 7, in import data\nModule Not Found Error: No module named ‘dataset'
\n\nHow can I fix it? Thanks.
", "votes": null, "replies": [ { "id": 237537, "author_name": "donghp", "author_url": "", "post_date": "10/30/2017 14:41:52", "content": "I clear it.
", "votes": null, "replies": [] } ] }, { "id": 246079, "author_name": "vasaibaby", "author_url": "", "post_date": "11/20/2017 14:27:25", "content": "Hi folks,
\n\nI am facing a strange error where .py files are getting executed until the tensorflow throws the FileNotFoundError : tensorflow.python.framework.errors_impl.NotFoundError: NewRandomAccessFile failed to Create/Open: frozen/inception/classify_image_graph_def.pb : The system cannot find the path specified`
My guess is : setup is missing from the path since I'm calling out from root directory. Is this correct?
Any idea on how to resolve this?
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "235538": "I have played with dogs classification. Here some practical results - https://medium.com/@kirillpanarin/dog-breed-classification-hands-on-approach-b5e4f88c333e", "235924": "Thanks for the code, Kirill. I am trying to read it through and on this stage `python -m src.data_preparation.stanford_ds_to_tfrecords` it looks like `one_hot_encoder, _ = dataset.one_hot_label_encoder()` expects labels already to be prepared and saved in `paths.LABELS`, but I cant find a piece of code which should do that.", "236025": "Hi SM! Thanks for letting me know about the issue. The code expected labels.csv from this competition training dataset. I didn't check it in. I moved the list of all breeds to the new file 'data/breeds.csv' and modified all code to use it instead of loading 'data/train/labels.csv' and taking unique values of breeds. Please, pull the latest changes from the repo and give it a try again - it has to work now.", "237477": "", "237484": "Thanks for good writing, Kirill. \nI have a question though.\nWhen I execute\n‘python -m src.data_preparation.stanford_ds_to_tfrecords’ in the root of your repo.\nI got a warning\nFile “C:\\Users\\ME\\dog-breeds-classification\\src\\data_preparation\\stanford_ds_to_tfrecords.py”, line 7, in", "237537": "I clear it.", "246079": "Hi folks,\n\nI am facing a strange error where `.py` files are getting executed until the tensorflow throws the `FileNotFoundError` : `tensorflow.python.framework.errors_impl.NotFoundError: NewRandomAccessFile failed to Create/Open: frozen/inception/classify_image_graph_def.pb` : The system cannot find the path specified` \n\nMy guess is : `setup` is missing from the `path` since I'm calling out from root directory. Is this correct?\n\n\nAny idea on how to resolve this?", "246128": "Hi SM, could you help me understand the `tf_record_utils` in the `stanford_ds_to_tfrecords.py` code. I get a module not found error for this one. When trying to install this module using `pip` I get `No matching distribution found for tf_record_utils` error", "246134": "Its a local py file which you can find here dog-breeds-classification/src/data_preparation/tf_record_utils.py", "246173": "got it.. So it should be found automatically right when I run the `stanford_ds_to_tfrecords.py` file. But somehow, its not able to find this.\n\nIn which directory am i suppose to run the conversion file? inside the `src` folder ? I guess, I need to add some folders to the `PATH` variables", "246184": "Hi Joel. \nIn 'stanford_ds_to_tfrecords.py' change \n(line number: 12) from tf_record_utils import * -> from src.data_preparation import tf_record_utils \nI hope it helps.", "246660": "Hi DonghunPark, thank you for helping out. It worked for me!\nCould you look into the error i'm facing now(raised it in a separate comment?)" }, "source": "meta" }