The Large Scale Movie Description Challenge (LSMDC) 2017 : Movie Description Forum

Go back to competition Back to thread list Post in this thread

> Using Auxiliary Dataset in Training

1. Can I use an auxiliary dataset in the training? For example, if I have access to another dataset of video shots (taken from movies for example) and their corresponding descriptions (ground truth). Can I use this dataset? Or we are only allowed to use the official datasets of the competition?
2. can I train my language model with more vocabulary that does not exist in the sentences of the provided datasets (M-VAD, MP-2)?

Posted by: noureldien @ Feb. 1, 2016, 8:26 p.m.

I received this reply from the organizer:
1. Yes, it is allowed. You will be asked to specify the details for your additional training data during the submission to the codalab server.
2. The challenge will keep running and we plan to add another track to it. As of now I can not tell whether we will organize a workshop at ECCV'16 or not. This will become clear by the end of February.

Posted by: noureldien @ Feb. 5, 2016, 9:06 a.m.
Post in this thread