How to implement a caffe net with multiple images match one label?

[Zhenhua Chen]

How to implement a caffe net with multiple images match one label? I am doing experiment on videos, so I just wonder if this is possible using current caffe. Thank you!

[Jan C Peters]

If by "multiple" you mean a fixed number that is the same for each label, then it is certainly possible. If this number is not fixed, you are likely dealing with a sequence learning task, and standard caffe is not very suited for that afaik. But there are some forks of caffe implementing recurrency and LSTMs, these might be interesting to you.

Jan

[Zhenhua Chen]

Hi Jan,
Thank you very much for replying! I actually figure out a way to do this by separating the label and the data, and then slice, reshape and concatenate the data. However, maybe RNN is a better idea, I will try that. Again, thanks for your advice!

Zhenhua