$ \newcommand{\cc}[1]{\color{black}{#1}} \newcommand{\mvec}[1]{\mathbf{#1}} \newcommand{\cvec}[2]{^\mathrm{#2}\mathbf{#1}} \newcommand{\ctrans}[3]{^\mathrm{#3}\mathbf{#1}_\mathrm{#2}} \newcommand{\rmat}[9]{\cc{\begin{bmatrix}{#1}&{#2}&{#3}\\\ {#4}&{#5}&{#6}\\\ {#7}&{#8}&{#9}\end{bmatrix}}} \newcommand{\hmat}[3]{\cc{\begin{bmatrix}{#1}\\\ {#2}\\\ {#3}\\\ \end{bmatrix}}} \newcommand{\mq}[4]{\cc{\begin{bmatrix}{#1}&{#2}&{#3}&{#4} \end{bmatrix}}} \newcommand{\nvec}[3]{\cc{\begin{bmatrix}{#1}&{#2}&{#3} \end{bmatrix}}} \newcommand{\vvec}[3]{\cc{\begin{bmatrix}{#1}\\\ {#2}\\\ {#3}\end{bmatrix}}} \newcommand{\vvt}[2]{\cc{\begin{bmatrix}{#1}\\\ {#2}\end{bmatrix}}} \newcommand{\calmat}[4]{\cc{\begin{bmatrix}{#1}&0&{#3}\\\ 0&{#2}&{#4}\end{bmatrix}}} \newcommand{\dp}{^{\prime\prime}} $

Download

Preparing for Download

The installation instructions and scripts have been tested on Ubuntu 14.04. Porting the scripts to any unix-like operating system should be straight forward. The Windows OS is not supported, but we provide raw download links so you can piece the data together yourself.

The full data set, without the frames extracted, has a size of about 25GB. If frames are extracted for all cameras and all sequences, the total size balloons to more than 100GB, so having some head room is definitely a good idea. For most purposes, 50GB should be more than enough space.

Running the Installer

The install scripts are in the “master” branch of the github repository, and can be run like this (you may need to install git if it’s not on your machine yet, using sudo apt-get install git):

git clone https://github.com/daniilidis-group/penncosyvio.git
cd penncosyvio/tools
./download.bash 
================ PENNCOSYVIO INSTALLER ======================
------------ select sensor to download:
1) GoPro C1		  4) Tango Bottom	    7) Intrinsic Calibration
2) GoPro C2		  5) Tango Top		    8) Extrinsic Calibration
3) GoPro C3		  6) VI Sensor		    9) Quit
enter your choice: 

Pick the sensors that you are interested in and download the data, which will end up in the data subdirectory. Once you quit out of the installer tool, change into the data directory and see what’s there. This is how it looks if you downloaded only the Tango Top data:

cd ..
tree data
data
└── tango_top
    ├── af
    │   ├── accelerometer.txt
    │   ├── gyroscope.txt
    │   ├── seq_af.mp4
    │   └── timestamps.txt
    ├── as
    │   ├── accelerometer.txt
    │   ├── gyroscope.txt
    │   ├── seq_as.mp4
    │   └── timestamps.txt
    ├── bf
    │   ├── accelerometer.txt
    │   ├── gyroscope.txt
    │   ├── seq_bf.mp4
    │   └── timestamps.txt
    └── bs
        ├── accelerometer.txt
        ├── gyroscope.txt
        ├── seq_bs.mp4
        └── timestamps.txt

Running the Frame Extractor

Because reading mp4 files programmatically can be difficult (and notably because Matlab on Ubuntu skips frames due to some issues with the gstreamer framework), we provide a bash script to extract frames in PNG format. The script relies on either ffmpeg or avconv as an extraction tool. Either of them must be installed beforehand, avconf being easier to install as it is part of the Ubuntu distibution. Here is how to use the extraction tool:

cd tools
./extract_frames.bash
------------ select frames to extract:
1) GoPro C1	 3) GoPro C3	  5) Tango Top
2) GoPro C2	 4) Tango Bottom  6) Quit
enter your choice: 5
------------- unpacking frames for Tango Top
 Be patient, this will take about 10-15min/sequence!
....

The extracted frames will end up in the respective subdirectory of the data subdirectory.

Manual Download

Here are links for direct download of the data files:

After download, untar them in the data directory.