"No input data" - Improve README to run MaskFusion

[NicksonYap]

Hi,

Finally managed to get MaskFusion to open.

Turns out Kinect v1 & v2 is supported, althrough v2 requries a additional installation of libfreenect2

Asus Xtion Live (with RGB) is also supported and taken as v1

To run Asus Xtion Live or Kinect v1: (tested with Xtion)

./MaskFusion -run -v1

To run Kinect v2: (need to set WITH_FREENECT2 during build) (not tested yet)

./MaskFusion -run -v2

[NicksonYap]

I might send a PR, especially noting that only up to Ubuntu 16.04 is supported

Tried Ubuntu 18.04 and gave a real headache as CUDA 10 is not actually working for latest Tensorflow (1.12.0) but recommended for Ubuntu 18.04

Ubuntu 16.04 and TF 1.12.0 requires CUDA 9.0 but 8.0 was installed by build.sh instead

How to Install CUDA 9.0 for TF: https://www.tensorflow.org/install/gpu#install_cuda_with_apt

[archenroot]

@NicksonYap - thank you for this review, you could update README file with these sensor details... I think out of all these Kinect v2 is having best capabilities overall:
http:https://docs.ipisoft.com/Depth_Sensors_Comparison