adding "stream" mode to VideoDataLayer

[Efreeto]

The "stream" mode lets the users extract features from videos without having to specify the frame numbers in the input or output text files. It can also load from live camera. The original behaviour to load videos is kept intact, and also any modifications are welcome.

How to use:

1. Set the new flag 'use_stream' to true in the video data layer in the prototxt file.
2. Change "extract_image_features.bin" argument 'num_mini_batches' to a negative number. (to indicate an infinite number)
3. Change the input text file format
   ex) input/avi/v_ApplyEyeMakeup_g01_c01.avi 0 16 0
   [filename] [start_frm] [interval] [label]
   You can put 'CAMERA' in [filename] to load from the live camera.
4. Change the output text file format
   ex) output/c3d/
   In the stream mode, the output location is automatically determined from the input, as follows [output_location][input_filename][start_frm].txt. So only the output_location is needed.

• I provided an example script "c3d_sport1m_feature_extraction_video**_stream**.sh" to demonstrate the usage.

[ghost]

Thank you for your pull request and welcome to our community. We require contributors to sign our Contributor License Agreement, and we don't seem to have you on file. In order for us to review and merge your code, please sign up at https://code.facebook.com/cla - and if you have received this in error or have any questions, please drop us a line at [email protected]. Thanks!

[ghost]

Thank you for signing our Contributor License Agreement. We can now accept your code for this (and any) Facebook open source project. Thanks!