Are there any restrictions regarding the method? Can we also do an offline recognition of the actions or only online?
Created by Stefanie Krell Steffii1223 @LiYC, if you do not wish to open a new forum thread, please email us at endovis-weiss-vision@ucl.ac.uk and make sure you add the tag [RARP50] at the beginning of your email's subject. I'm also a participant in the Action Recognition Challenge, and I'd like to talk to you about model training. May I have your email?
Thank you for the clarification. We allow both online and offline methods By offline I meant methods that can look at the whole video (so past and future frames) for each frame prediction.
And with online I meant methods that process only past frames for each frame prediction.
I hope that clarifies my question.
Hi, we require participants to submit fully automatic methods, meaning that a method should process files, stored under a predetermined file structure without any user input.
Can you please clarify what you refer to with the tearms offline and online? If you refer to the runtime performance of methods ( their ability to predict in real time) we do not have any special requirements.