Dear all,
\nwe have been changed a little bit the time-line, in order to clarify the steps.
\nOn August the 1st, the Test data and the labeled validation data will be published. In the case of Test data, it will be encrypted, therefore, you can download it but you need the decryption key to open it. You have 10 days to download the data and to start training your methods with the full training data (Training + Validation).
\nOn August the 10th, the decryption key will be made available, so you can unzip the final evaluation data (Test) and run your methods in order to submit both, the final code and the predictions, to the Kaggle platform. The deadline is the 20th of August.
\nFinally, the teams should send before August the 25th the fact sheets, explaining their approach and used methods.
\nRegarding to the submitted code
\nSubmitted code can be written in any language, but it should have a main method to run the recognition process. A Readme file should be provided explaining how to run the code and providing all information about system requirements/dependencies, to facilitate the code verification.
\nRegarding to the fact sheets
\nWe do not restrict the format of fact sheets, a PowerPoint or equivalent presentation with 2 or 3 slides should be enough. Nevertheless, they must contain the names and affiliation of the team, a general overview of the approach and the used methods.
\nRegarding the predictions
\nThe final predictions will be submitted in the same format than the predictions you are submitting for the validation data.
\n\n
I hope that this information will clarify the process. Do not hesitate to ask any additional information.
\nXavier
", "messages": [ { "id": "27475", "postDate": "07/20/2013 15:47:58", "content": "Dear all,
\nwe have been changed a little bit the time-line, in order to clarify the steps.
\nOn August the 1st, the Test data and the labeled validation data will be published. In the case of Test data, it will be encrypted, therefore, you can download it but you need the decryption key to open it. You have 10 days to download the data and to start training your methods with the full training data (Training + Validation).
\nOn August the 10th, the decryption key will be made available, so you can unzip the final evaluation data (Test) and run your methods in order to submit both, the final code and the predictions, to the Kaggle platform. The deadline is the 20th of August.
\nFinally, the teams should send before August the 25th the fact sheets, explaining their approach and used methods.
\nRegarding to the submitted code
\nSubmitted code can be written in any language, but it should have a main method to run the recognition process. A Readme file should be provided explaining how to run the code and providing all information about system requirements/dependencies, to facilitate the code verification.
\nRegarding to the fact sheets
\nWe do not restrict the format of fact sheets, a PowerPoint or equivalent presentation with 2 or 3 slides should be enough. Nevertheless, they must contain the names and affiliation of the team, a general overview of the approach and the used methods.
\nRegarding the predictions
\nThe final predictions will be submitted in the same format than the predictions you are submitting for the validation data.
\n\n
I hope that this information will clarify the process. Do not hesitate to ask any additional information.
\nXavier
", "rawMarkdown": "", "votes": null }, { "id": "27634", "postDate": "07/25/2013 01:05:01", "content": "Dear Xavier,
\n\n
Thank you for organizing this competition!
\nFrom your above post, we receive the decrypt key for the final evaluation data on Aug 10, and the deadline for submitting code+prediction-on-final-eval-data on Aug 20. Hence we have 10 days to work on the final eval data.
\nHowever, on the other main page ( http://gesture.chalearn.org/mmschedule ), the schedule is different. I reproduce it here (correct as at my viewing on 25-July 9:00 GMT+8):
\nMay 25, 2013: Full release of development data. Validation data will be released on June 3.
\nAugust 1, 2013: Release of final evaluation data and validation ground truth labels.
\nAugust 10, 2013: Release of final evaluation data decryption key.
\nAugust 15, 2013: End of the quantitative competition. Deadline for code submission and the prediction results on final evaluation data. The organizers start the code verification by running it on the final evaluation data.
\nAugust 25, 2013: Deadline for submitting the fact sheets summarizing the proposed methods.
September 1, 2013: Release of the verification results to the participants for review
\n
Note that in the above, the deadline for code+prediction submission is Aug 15. This can be confusing. Can you do an update?
\n\n
Another question regarding the code. Are we expected to submit code to produce the prediction on the final eval data? Or are we expected to also submit code for feature extraction, model learning, etc ?
\nIf it is the latter, our code can be fairly complicated as it involves the installation of a few modules, and setting up the runtime environment. In the event that our code runs into problem on your verification platform, will you be contacting us ?
\nBasically, I understand that as organizers, you want to ensure that the prediction is done by machine computation, and not by manual labeling. The best way of ensuring this, is to have a set of secret video files that you have reserved for verification, and undisclosed to all competition participants. In so doing, then you would require that our code perform feature extraction and model learning.
\n\n
KW
\ntelepoints
\n\n
", "rawMarkdown": "", "votes": null } ], "comments": [ { "id": 27634, "author_name": "kongwah", "author_url": "", "post_date": "07/25/2013 01:05:01", "content": "
Dear Xavier,
\n\n
Thank you for organizing this competition!
\nFrom your above post, we receive the decrypt key for the final evaluation data on Aug 10, and the deadline for submitting code+prediction-on-final-eval-data on Aug 20. Hence we have 10 days to work on the final eval data.
\nHowever, on the other main page ( http://gesture.chalearn.org/mmschedule ), the schedule is different. I reproduce it here (correct as at my viewing on 25-July 9:00 GMT+8):
\nMay 25, 2013: Full release of development data. Validation data will be released on June 3.
\nAugust 1, 2013: Release of final evaluation data and validation ground truth labels.
\nAugust 10, 2013: Release of final evaluation data decryption key.
\nAugust 15, 2013: End of the quantitative competition. Deadline for code submission and the prediction results on final evaluation data. The organizers start the code verification by running it on the final evaluation data.
\nAugust 25, 2013: Deadline for submitting the fact sheets summarizing the proposed methods.
September 1, 2013: Release of the verification results to the participants for review
\n
Note that in the above, the deadline for code+prediction submission is Aug 15. This can be confusing. Can you do an update?
\n\n
Another question regarding the code. Are we expected to submit code to produce the prediction on the final eval data? Or are we expected to also submit code for feature extraction, model learning, etc ?
\nIf it is the latter, our code can be fairly complicated as it involves the installation of a few modules, and setting up the runtime environment. In the event that our code runs into problem on your verification platform, will you be contacting us ?
\nBasically, I understand that as organizers, you want to ensure that the prediction is done by machine computation, and not by manual labeling. The best way of ensuring this, is to have a set of secret video files that you have reserved for verification, and undisclosed to all competition participants. In so doing, then you would require that our code perform feature extraction and model learning.
\n\n
KW
\ntelepoints
\n\n
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "27475": "", "27634": "" }, "source": "meta" }