Upgrade to Tensorflow 1.4 Remove base images instead have a single container image.

remove CRFasRNN.

Next:
1. Create example clients for retrieval experiments.
2. Ability to import region annotations specified in JSON file with filenames and frame_index used as invariant identifiers.
3. Streamlining creation of pre-indexed datasets.
4. Reduce use of dvaapp in dvaui. Eventually ensuring that everything goes through the REST API.
5. Store uploaded videos/zip files to "ingest"  directory and use create_launch functionality to separate interface cleanly and ensuring everything goes through DVAPQL.
6. improve paper & documentation

Author: akshay bhat

deploy/dockerfiles/Dockerfile

@@ -29,7 +29,6 @@ RUN pip install --no-cache-dir http://download.pytorch.org/whl/cu75/torch-0.1.12
 RUN pip install --no-cache-dir mxnet==0.11.0
 RUN pip install torchvision
 WORKDIR "/root/DVA"
-RUN cd dvalib/crfasrnn/cpp && ./compile.sh && cd ../../..
 WORKDIR "/root/DVA/repos/lopq/python"
 RUN python setup.py install
 WORKDIR "/root/DVA"