{ "id": 216653, "title": "Full Case study of cassava leaf challenge[feedback needed]", "url": "/competitions/cassava-leaf-disease-classification/discussion/216653", "author_name": "", "post_date": "2021-02-03T14:49:58.500539900Z", "votes": 2, "comment_count": 3, "views": 0, "content": "

Hello,
\n I have documented my entire cassava challenge journey in this Notebook.

\n

Please have a look and give me some feedback. No need for upvotes, just genuine feedback on what else to add, the flow of the approach, etc would make my day.

\n

Thank you very much.

\n

P.S: The current version has an OOM error in the training phase just ignore it.

", "messages": [ { "id": "1184490", "postDate": "02/03/2021 14:49:58", "content": "

Hello,
\n I have documented my entire cassava challenge journey in this Notebook.

\n

Please have a look and give me some feedback. No need for upvotes, just genuine feedback on what else to add, the flow of the approach, etc would make my day.

\n

Thank you very much.

\n

P.S: The current version has an OOM error in the training phase just ignore it.

", "rawMarkdown": "Hello,\n I have documented my entire cassava challenge journey in this [Notebook](https://www.kaggle.com/mohneesh7/cassava-2020-case-study-full-progress-keras-gpu).\n\nPlease have a look and give me some feedback. No need for upvotes, just genuine feedback on what else to add, the flow of the approach, etc would make my day.\n\nThank you very much.\n\nP.S: The current version has an OOM error in the training phase just ignore it.", "votes": null }, { "id": "1194793", "postDate": "02/10/2021 11:24:32", "content": "

Very good notebook I like it, I would suggest to increase the minority classes and fill the gap between classes with different methods implemented in opencv; Pyramids, Geometric transformation, Smoothing images, and Image gradients.
\nAfter balancing the dataset, you can use it to find more insights using opencv like haar and hog features, thus this is an optional step.
\nYou can also try more pre-trained models: Xception and Inception.
\nVery good and honest work, good luck doing more and better projects

", "rawMarkdown": "Very good notebook I like it, I would suggest to increase the minority classes and fill the gap between classes with different methods implemented in opencv; Pyramids, Geometric transformation, Smoothing images, and Image gradients.\nAfter balancing the dataset, you can use it to find more insights using opencv like haar and hog features, thus this is an optional step.\nYou can also try more pre-trained models: Xception and Inception.\nVery good and honest work, good luck doing more and better projects", "votes": null }, { "id": "1194947", "postDate": "02/10/2021 13:23:41", "content": "

Thank you very much, man. I really appreciate it.
\nWill try to better me in the future.

", "rawMarkdown": "Thank you very much, man. I really appreciate it.\nWill try to better me in the future.", "votes": null }, { "id": "1195058", "postDate": "02/10/2021 14:28:58", "content": "

Try to be better of course that's important, but it doesn't mean this is bad or even average. I see that as a very honest work, keep up the good work

", "rawMarkdown": "Try to be better of course that's important, but it doesn't mean this is bad or even average. I see that as a very honest work, keep up the good work", "votes": null } ], "comments": [ { "id": 1194793, "author_name": "omarmohamed22", "author_url": "", "post_date": "02/10/2021 11:24:32", "content": "

Very good notebook I like it, I would suggest to increase the minority classes and fill the gap between classes with different methods implemented in opencv; Pyramids, Geometric transformation, Smoothing images, and Image gradients.
\nAfter balancing the dataset, you can use it to find more insights using opencv like haar and hog features, thus this is an optional step.
\nYou can also try more pre-trained models: Xception and Inception.
\nVery good and honest work, good luck doing more and better projects

", "votes": null, "replies": [ { "id": 1194947, "author_name": "mohneesh7", "author_url": "", "post_date": "02/10/2021 13:23:41", "content": "

Thank you very much, man. I really appreciate it.
\nWill try to better me in the future.

", "votes": null, "replies": [] }, { "id": 1195058, "author_name": "omarmohamed22", "author_url": "", "post_date": "02/10/2021 14:28:58", "content": "

Try to be better of course that's important, but it doesn't mean this is bad or even average. I see that as a very honest work, keep up the good work

", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "1184490": "Hello,\n I have documented my entire cassava challenge journey in this [Notebook](https://www.kaggle.com/mohneesh7/cassava-2020-case-study-full-progress-keras-gpu).\n\nPlease have a look and give me some feedback. No need for upvotes, just genuine feedback on what else to add, the flow of the approach, etc would make my day.\n\nThank you very much.\n\nP.S: The current version has an OOM error in the training phase just ignore it.", "1194793": "Very good notebook I like it, I would suggest to increase the minority classes and fill the gap between classes with different methods implemented in opencv; Pyramids, Geometric transformation, Smoothing images, and Image gradients.\nAfter balancing the dataset, you can use it to find more insights using opencv like haar and hog features, thus this is an optional step.\nYou can also try more pre-trained models: Xception and Inception.\nVery good and honest work, good luck doing more and better projects", "1194947": "Thank you very much, man. I really appreciate it.\nWill try to better me in the future.", "1195058": "Try to be better of course that's important, but it doesn't mean this is bad or even average. I see that as a very honest work, keep up the good work" }, "source": "meta" }