Use Caffe2 base docker image from the docker hub

Reviewed By: rbgirshick

Differential Revision: D6824147

fbshipit-source-id: fe06c4b4ee200003d5d9e4ca62e4afc9741cd929

INSTALL.md

@@ -106,28 +106,19 @@ python2 $DETECTRON/tests/test_zero_even_op.py
 
 ## Docker Image
 
-We provide a [`Dockerfile`](docker/Dockerfile) that you can use to build a Detectron image on top of a Caffe2 image that satisfies the requirements outlined at the top. If you're using a prebuilt Caffe2 image (e.g. from the [docker repo](https://hub.docker.com/r/caffe2ai/caffe2/)), please make sure that it includes the [Detectron module](https://github.com/caffe2/caffe2/tree/master/modules/detectron). We also provide an example of how to build an up-to-date Caffe2 image.
+We provide a [`Dockerfile`](docker/Dockerfile) that you can use to build a Detectron image on top of a Caffe2 image that satisfies the requirements outlined at the top. If you would like to use a Caffe2 image different from the one we use by default, please make sure that it includes the [Detectron module](https://github.com/caffe2/caffe2/tree/master/modules/detectron).
 
-Build a Caffe2 image:
-
-```
-cd /path/to/caffe2/docker/ubuntu-16.04-cuda8-cudnn6-all-options
-# Use the latest Caffe2 master
-sed -i -e 's/ --branch v0.8.1//g' Dockerfile
-docker build -t caffe2:cuda8-cudnn6-all-options .
-```
-
-Build a Detectron image:
+Build the image:
 
 ```
 cd $DETECTRON/docker
-docker build -t detectron:c2-cuda8-cudnn6 .
+docker build -t detectron:c2-cuda9-cudnn7 .
 ```
 
-Run the Detectron image (e.g. for [`BatchPermutationOp test`](tests/test_batch_permutation_op.py)):
+Run the image (e.g. for [`BatchPermutationOp test`](tests/test_batch_permutation_op.py)):
 
 ```
-nvidia-docker run --rm -it detectron:c2-cuda8-cudnn6 python2 tests/test_batch_permutation_op.py
+nvidia-docker run --rm -it detectron:c2-cuda9-cudnn7 python2 tests/test_batch_permutation_op.py
 ```
 
 ## Troubleshooting

docker/Dockerfile

 # Use Caffe2 image as parent image
-FROM caffe2:cuda8-cudnn6-all-options
+FROM caffe2/caffe2:snapshot-py2-cuda9.0-cudnn7-ubuntu16.04
+
+RUN mv /usr/local/caffe2 /usr/local/caffe2_build
+ENV Caffe2_DIR /usr/local/caffe2_build
+
+ENV PYTHONPATH /usr/local/caffe2_build:${PYTHONPATH}
+ENV LD_LIBRARY_PATH /usr/local/caffe2_build/lib:${LD_LIBRARY_PATH}
 
 # Install Python dependencies
 RUN pip install numpy>=1.13 pyyaml>=3.12 matplotlib opencv-python>=3.2 setuptools Cython mock
 
 # Install the COCO API
-RUN git clone https://github.com/cocodataset/cocoapi.git
+RUN git clone https://github.com/cocodataset/cocoapi.git /cocoapi
 WORKDIR /cocoapi/PythonAPI
 RUN make install
 

lib/CMakeLists.txt

@@ -26,11 +26,17 @@ add_library(
      caffe2_detectron_custom_ops SHARED
      ${CUSTOM_OPS_CPU_SRCS})
 
+target_include_directories(
+    caffe2_detectron_custom_ops PRIVATE
+    ${CAFFE2_INCLUDE_DIRS})
 target_link_libraries(caffe2_detectron_custom_ops caffe2)
 install(TARGETS caffe2_detectron_custom_ops DESTINATION lib)
 
 # Install custom GPU ops lib.
 if (HAVE_CUDA)
+  # Additional -I prefix is required for CMake versions before commit (< 3.7):
+  # https://github.com/Kitware/CMake/commit/7ded655f7ba82ea72a82d0555449f2df5ef38594
+  list(APPEND CUDA_INCLUDE_DIRS -I${CAFFE2_INCLUDE_DIRS})
   CUDA_ADD_LIBRARY(
       caffe2_detectron_custom_ops_gpu SHARED
       ${CUSTOM_OPS_CPU_SRCS}