A Tensorflow implementation of the Deep Canonical Time Warping.

Deep Canonical Time Warping
G. Trigeorgis, M. A. Nicolaou, S. Zafeiriou, B. Schuller.
Proceedings of IEEE International Conference on Computer Vision & Pattern Recognition (CVPR'16).
June 2016.

We are an avid supporter of the Menpo project (http://www.menpo.org/) which we use in various ways throughout the implementation.

In general, as explained in Menpo's installation instructions, it is highly recommended to use conda as your Python distribution.

Once downloading and installing conda, this project can be installed by:

Step 1: Create a new conda environment and activate it:

$ conda create -n dctw python=3.5
$ source activate dctw

Step 2: Install TensorFlow following the official installation instructions. For example, for 64-bit Linux, the installation of CPU-only, Python 3.5 TensorFlow involves:

(dctw)$ pip install --upgrade https://storage.googleapis.com/tensorflow/linux/cpu/tensorflow-0.12.0rc0-cp35-cp35m-linux_x86_64.whl

Step 3: Install menpo from the menpo channel as:

(dctw)$ conda install -c menpo menpo

Step 4: Compile the extra TraceNormOp op.

(dctw)$ bash compile_cpu.sh

When you are done you can go through the example in demo.ipynb.