{ "id": 395986, "title": "Smaller Dataset Size Request ", "url": "/competitions/asl-signs/discussion/395986", "author_name": "", "post_date": "2023-03-19T19:52:29.831373100Z", "votes": 1, "comment_count": 4, "views": 0, "content": "

I will be grateful if I the dataset for this competition is crunched by someone for me and others so we can participate. Please do tag me when you do so. Thank you in advance.

", "messages": [ { "id": "2188606", "postDate": "03/19/2023 19:52:29", "content": "

I will be grateful if I the dataset for this competition is crunched by someone for me and others so we can participate. Please do tag me when you do so. Thank you in advance.

", "rawMarkdown": "I will be grateful if I the dataset for this competition is crunched by someone for me and others so we can participate. Please do tag me when you do so. Thank you in advance.", "votes": null }, { "id": "2188985", "postDate": "03/20/2023 06:26:10", "content": "

You could find ready-to-use reduced datasets in the works, like here

\n

There are a bunch of other preprocessed datasets (like a mean of each landmark in time in .npy format like resized to 3 frames suqences in the .tfrecord format)

\n

I hope, that would help

", "rawMarkdown": "You could find ready-to-use reduced datasets in the works, like [here](https://www.kaggle.com/code/roberthatch/gislr-lb-0-63-on-the-shoulders#PREPROCESSINGrl)\n\nThere are a bunch of other preprocessed datasets (like a mean of each landmark in time in .npy format like [resized to 3 frames suqences in the .tfrecord format](https://www.kaggle.com/code/lonnieqin/islr-create-tfrecord))\n\nI hope, that would help", "votes": null }, { "id": "2189246", "postDate": "03/20/2023 10:51:51", "content": "

It will surely help. Thank you very much @meowmeowmeowmeowmeow

", "rawMarkdown": "It will surely help. Thank you very much @meowmeowmeowmeowmeow", "votes": null }, { "id": "2191194", "postDate": "03/21/2023 19:38:06", "content": "

@lordxerxes here are the most popular one:

\n\n

Hope that helps!

", "rawMarkdown": "lordxerxes here are the most popular one:\n- [npy dataset](https://www.kaggle.com/code/mayukh18/gislr-feature-data) with mean and std frame for each file (one row of 543x3x2=3258 columns for each file)\n- [npy dataset](https://www.kaggle.com/datasets/lonnieqin/isolated-sign-language-aggregation-dataset) with frames aggregated (1 frame for each file of shape (543 rows, 3 columns)\n- [TFRecords](https://www.kaggle.com/code/lonnieqin/islr-create-tfrecord) with fixed frames (12)\n- [TFRecords](https://www.kaggle.com/code/josephzahar/tfrecords-type-asl) split by type (right hand, left hand, pose, face) \n\nHope that helps!", "votes": null }, { "id": "2191281", "postDate": "03/21/2023 21:01:15", "content": "

Surely, it does. Thanks a lot @josephzahar

", "rawMarkdown": "Surely, it does. Thanks a lot @josephzahar", "votes": null } ], "comments": [ { "id": 2188985, "author_name": "meowmeowmeowmeowmeow", "author_url": "", "post_date": "03/20/2023 06:26:10", "content": "

You could find ready-to-use reduced datasets in the works, like here

\n

There are a bunch of other preprocessed datasets (like a mean of each landmark in time in .npy format like resized to 3 frames suqences in the .tfrecord format)

\n

I hope, that would help

", "votes": null, "replies": [ { "id": 2189246, "author_name": "lordxerxes", "author_url": "", "post_date": "03/20/2023 10:51:51", "content": "

It will surely help. Thank you very much @meowmeowmeowmeowmeow

", "votes": null, "replies": [] } ] }, { "id": 2191194, "author_name": "josephzahar", "author_url": "", "post_date": "03/21/2023 19:38:06", "content": "

@lordxerxes here are the most popular one:

\n\n

Hope that helps!

", "votes": null, "replies": [ { "id": 2191281, "author_name": "lordxerxes", "author_url": "", "post_date": "03/21/2023 21:01:15", "content": "

Surely, it does. Thanks a lot @josephzahar

", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "2188606": "I will be grateful if I the dataset for this competition is crunched by someone for me and others so we can participate. Please do tag me when you do so. Thank you in advance.", "2188985": "You could find ready-to-use reduced datasets in the works, like [here](https://www.kaggle.com/code/roberthatch/gislr-lb-0-63-on-the-shoulders#PREPROCESSINGrl)\n\nThere are a bunch of other preprocessed datasets (like a mean of each landmark in time in .npy format like [resized to 3 frames suqences in the .tfrecord format](https://www.kaggle.com/code/lonnieqin/islr-create-tfrecord))\n\nI hope, that would help", "2189246": "It will surely help. Thank you very much @meowmeowmeowmeowmeow", "2191194": "lordxerxes here are the most popular one:\n- [npy dataset](https://www.kaggle.com/code/mayukh18/gislr-feature-data) with mean and std frame for each file (one row of 543x3x2=3258 columns for each file)\n- [npy dataset](https://www.kaggle.com/datasets/lonnieqin/isolated-sign-language-aggregation-dataset) with frames aggregated (1 frame for each file of shape (543 rows, 3 columns)\n- [TFRecords](https://www.kaggle.com/code/lonnieqin/islr-create-tfrecord) with fixed frames (12)\n- [TFRecords](https://www.kaggle.com/code/josephzahar/tfrecords-type-asl) split by type (right hand, left hand, pose, face) \n\nHope that helps!", "2191281": "Surely, it does. Thanks a lot @josephzahar" }, "source": "meta" }