Hello, I'm new here at Kaggle and I have a basic question regarding to this:
\n\"In this competition you will be submitting a TensorFlow Lite model file. The model must take one or more landmark frames as an input and return a float vector (the predicted probabilities of each sign class) as the output\"
\nAs I understand it, the only way to submit your solution is through a Tensorflow Lite model file.
\nDoes that mean then that I can't have a solution that is a Python program that uses my trained model to analyze the frames and later analyze the results of each frame and generate an output?
\nAre all competitions like this?
Thats right. In this competition, you cannot submit a regular python code- only a TFLite. And all pre-processing and post-processing should be done inside and as part of the TFlight model. And, no, not all competitions are like that- this one is very rare. Usually, we submit regular code or even just predictions. Check the other competitions and see for yourself :)
", "rawMarkdown": "Thats right. In this competition, you cannot submit a regular python code- only a TFLite. And all pre-processing and post-processing should be done inside and as part of the TFlight model. And, no, not all competitions are like that- this one is very rare. Usually, we submit regular code or even just predictions. Check the other competitions and see for yourself :)", "votes": 2, "replies": [ { "id": 2303609, "postDate": "2023-06-15T11:34:47.010Z", "content": "Ok greySnow, Thank you!!
", "rawMarkdown": "Ok greySnow, Thank you!!" }, { "id": 2316951, "postDate": "2023-06-25T10:40:06.523Z", "content": "I'm honestly writing the python code along with the TFLite because I want to A. use it when I'm done and B. be able to test it on my end of things (so that means setting up video inputs, getting some form of openpose or maybe mediapipe going, etc.).
", "rawMarkdown": "I'm honestly writing the python code along with the TFLite because I want to A. use it when I'm done and B. be able to test it on my end of things (so that means setting up video inputs, getting some form of openpose or maybe mediapipe going, etc.)." } ] }, { "id": 2308045, "postDate": "2023-06-18T16:09:29.157Z", "content": "It's basically Google pushing their own framework, because they put up the prize money.
", "rawMarkdown": "It's basically Google pushing their own framework, because they put up the prize money. " }, { "id": 2302619, "postDate": "2023-06-14T16:51:57.503Z", "content": "Hello, I'm new here at Kaggle and I have a basic question regarding to this:
\n\"In this competition you will be submitting a TensorFlow Lite model file. The model must take one or more landmark frames as an input and return a float vector (the predicted probabilities of each sign class) as the output\"
\nAs I understand it, the only way to submit your solution is through a Tensorflow Lite model file.
\nDoes that mean then that I can't have a solution that is a Python program that uses my trained model to analyze the frames and later analyze the results of each frame and generate an output?
\nAre all competitions like this?
Thats right. In this competition, you cannot submit a regular python code- only a TFLite. And all pre-processing and post-processing should be done inside and as part of the TFlight model. And, no, not all competitions are like that- this one is very rare. Usually, we submit regular code or even just predictions. Check the other competitions and see for yourself :)
", "votes": 2, "replies": [ { "id": 2303609, "author_name": "Vostras", "author_url": "", "post_date": "2023-06-15T11:34:47.010000", "content": "Ok greySnow, Thank you!!
", "votes": 0, "replies": [] }, { "id": 2316951, "author_name": "Lars Renlund", "author_url": "", "post_date": "2023-06-25T10:40:06.523000", "content": "I'm honestly writing the python code along with the TFLite because I want to A. use it when I'm done and B. be able to test it on my end of things (so that means setting up video inputs, getting some form of openpose or maybe mediapipe going, etc.).
", "votes": 0, "replies": [] } ] }, { "id": 2308045, "author_name": "Robert Stratton", "author_url": "", "post_date": "2023-06-18T16:09:29.157000", "content": "It's basically Google pushing their own framework, because they put up the prize money.
", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "2302781": "Thats right. In this competition, you cannot submit a regular python code- only a TFLite. And all pre-processing and post-processing should be done inside and as part of the TFlight model. And, no, not all competitions are like that- this one is very rare. Usually, we submit regular code or even just predictions. Check the other competitions and see for yourself :)", "2308045": "It's basically Google pushing their own framework, because they put up the prize money. ", "2302619": "Hello, I'm new here at Kaggle and I have a basic question regarding to this:\n\"In this competition you will be submitting a TensorFlow Lite model file. The model must take one or more landmark frames as an input and return a float vector (the predicted probabilities of each sign class) as the output\"\nAs I understand it, the only way to submit your solution is through a Tensorflow Lite model file.\nDoes that mean then that I can't have a solution that is a Python program that uses my trained model to analyze the frames and later analyze the results of each frame and generate an output?\nAre all competitions like this?" } }