using Caffe dor video scenes classification

[caffe_user_vid]

Hi all,

 

I would like to discover whether is there a possibility classifying video scenes with caffe platform.

I know that Caffe is mainly oriented for images of any kind, though my main Q is if anyone's familiar with a possible adaptation for Caffe in order to classifying videos (each video is assembled from several frames, and frame number is various between videos). This adaptation should be compatible for the followed DNN layers ( conv , ReLU , Pool etc).

I noticed the "Concat" layer which may be involved in a video migration proccess of Caffe, though I still not sure how it's possible.

If it's not possible, can yiu please recommend any platform you think is compatible for my purposes.

Thanks,

[cooling]

Search this forum for a python data layer by Laura Hendricks. There's a github repo with an example of that exact thing.