Information
===========

This is the ShakeFive Dyadic Interaction dataset. The dataset consists of 100 videos 
showing both hand shakes and high fives.


==== Code ====

The dataset is accompanied by sample code that shows how to parse the Kinect XML metadata.
The sample code can be found in the "viewer" directory. It requires the c++11 standard and 
OpenCV to compile. Compilation can be done with CMake (see CMakeLists.txt). The code was
tested on Linux.

Main program entry in "viewer/ShakeFive.cpp"


==== Data ====

The image data can be found in the subdirectories the "data" directory, named "0000" to "0099".
Each of these directories contains the JPG images of the interaction sequence.

The data set can be downloaded from: http://www.staff.science.uu.nl/~0323217/shakefive/


==== XML ====

The XML metadata can be found in the "data" directory, named "0000.xml" to "0099.xml". The XML 
is structured as follows:

<video>
	<id/>								<!-- video id -->
	<image_path/>							<!-- path to frame images -->
	<video_size/>							<!-- image size -->
	<camera_matrix/>						<!-- camera matrix -->
	<dist_coeff/>							<!-- camera distortion matrix -->
	<frames_amount/>						<!-- amount of frames -->
	<frames>
		<frame_X>
			<id/>						<!-- frame id -->
			<image_filename/>				<!-- image filename for this frame -->
			<rotation/>					<!-- camera rotation matrix -->
			<translation/>					<!-- camera translation matrix -->
			<skeletons_amount/>				<!-- amount of skeletons in the frame -->
			<skeletons>
				<skeleton_X>
					<id/>				<!-- skeleton id -->
					<fid/>				<!-- frame id -->
					<aid/>				<!-- id of the action performed by this skeleton -->
					<action_name/>			<!-- name of the action performed by this skeleton -->
					<joints_amount/>		<!-- amount of joints for the skeleton -->
					<joints>
						<joint_X>
							<id/>		<!-- joint id -->
							<name/>		<!-- joint name/type -->
							<point2di/>	<!-- joint location in 2D -->
							<point3df/>	<!-- joint location in 3D -->
						</joint_X>
					</joints>
				</skeleton_X>
			</skeletons>
		</frame_X>
	<frames>
</video>


==== How to Cite ====

If you use this dataset, please cite:

  @misc{shakefive-r1,
    author       = "van Gemeren, C.J. and Tan, R. T. and Poppe, R. and Veltkamp, R. C.",
    title        = "Dyadic interaction detection from pose and flow",
    howpublished = "http://www.projects.science.uu.nl/shakefive/"
  }

