Hprc banner tamu.png

Difference between revisions of "SW:Faster-RCNN"

From TAMU HPRC
Jump to: navigation, search
(Faster-RCNN)
(Access)
Line 9: Line 9:
  
 
<font color=red>There are restrictions as to which version (GPU/CPU) of Faster-RCNN_TF works on each cluster. Please note these restrictions in the following sections and plan your jobs accordingly.</font>
 
<font color=red>There are restrictions as to which version (GPU/CPU) of Faster-RCNN_TF works on each cluster. Please note these restrictions in the following sections and plan your jobs accordingly.</font>
 +
 +
===Anaconda and Faster-RCNN_TF Packages===
 +
TAMU HPRC currently supports the user of Faster-RCNN_TF though the Anaconda modules. However, a user needs to install faster-rcnn_tf on their own following the instructions below:
 +
 +
[NetID@terra3 ~]$module load CUDA/8.0.44-GCC-system
 +
[NetID@terra3 ~]$module load Anaconda/2-5.0.1
 +
[NetID@terra3 ~]$source activate faster-r-cnn
 +
(faster-r-cnn)[NetID@terrra3 ~] now you need edit the lib/make.sh used for installation of faster-rcnn_tf as the following:
 +
 +
 +
TF_INC=$(python -c 'import tensorflow as tf; print(tf.sysconfig.get_include())')
 +
 +
CUDA_PATH=/sw/hprc/sw/Anaconda/2-5.0.1/envs/faster-r-cnn/
 +
CXXFLAGS=''
 +
HEADER='/sw/hprc/sw/Anaconda/2-5.0.1/envs/faster-r-cnn/lib/python2.7/site-packages/tensorflow/include/external/nsync/public'
 +
TF_LIB=$(python -c 'import tensorflow as tf; print(tf.sysconfig.get_lib())')
 +
 +
 +
if [[ "$OSTYPE" =~ ^darwin ]]; then
 +
CXXFLAGS+='-undefined dynamic_lookup'
 +
fi
 +
 +
cd roi_pooling_layer
 +
 +
if [ -d "$CUDA_PATH" ]; then
 +
nvcc -std=c++11 -c -o roi_pooling_op.cu.o roi_pooling_op_gpu.cu.cc \
 +
-I$TF_INC -I$HEADER -L$TF_LIB -D GOOGLE_CUDA=1 -x cu -Xcompiler -fPIC $CXXFLAGS --expt-relaxed-constexpr \
 +
-arch=sm_37
 +
 +
g++ -std=c++11 -shared -ltensorflow_framework -o roi_pooling.so roi_pooling_op.cc \
 +
roi_pooling_op.cu.o -I $TF_INC -L$TF_LIB -I $HEADER -D GOOGLE_CUDA=1 -fPIC $CXXFLAGS \
 +
-lcudart -L $CUDA_PATH/lib
 +
else
 +
g++ -std=c++11 -I$HEADER -L$TF_LIB -ltensorflow_framework -shared -o roi_pooling.so roi_pooling_op.cc \
 +
-I $TF_INC -fPIC $CXXFLAGS
 +
fi
 +
 +
cd ..
 +
 +
  
  
 
[[Category:Software]]
 
[[Category:Software]]

Revision as of 16:48, 12 February 2018

Faster-RCNN_TF

Description

This is an experimental Tensorflow implementation of Faster RCNN - a convnet for object detection with a region proposal network. For details about R-CNN please refer to the paper Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks by Shaoqing Ren, Kaiming He, Ross Girshick, Jian Sun.

Access

Faster-RCNN_TF is open to all HPRC users.

There are restrictions as to which version (GPU/CPU) of Faster-RCNN_TF works on each cluster. Please note these restrictions in the following sections and plan your jobs accordingly.

Anaconda and Faster-RCNN_TF Packages

TAMU HPRC currently supports the user of Faster-RCNN_TF though the Anaconda modules. However, a user needs to install faster-rcnn_tf on their own following the instructions below:

[NetID@terra3 ~]$module load CUDA/8.0.44-GCC-system
[NetID@terra3 ~]$module load Anaconda/2-5.0.1
[NetID@terra3 ~]$source activate faster-r-cnn
(faster-r-cnn)[NetID@terrra3 ~] now you need edit the lib/make.sh used for installation of faster-rcnn_tf as the following:


TF_INC=$(python -c 'import tensorflow as tf; print(tf.sysconfig.get_include())')

CUDA_PATH=/sw/hprc/sw/Anaconda/2-5.0.1/envs/faster-r-cnn/ CXXFLAGS= HEADER='/sw/hprc/sw/Anaconda/2-5.0.1/envs/faster-r-cnn/lib/python2.7/site-packages/tensorflow/include/external/nsync/public' TF_LIB=$(python -c 'import tensorflow as tf; print(tf.sysconfig.get_lib())')


if "$OSTYPE" =~ ^darwin ; then CXXFLAGS+='-undefined dynamic_lookup' fi

cd roi_pooling_layer

if [ -d "$CUDA_PATH" ]; then nvcc -std=c++11 -c -o roi_pooling_op.cu.o roi_pooling_op_gpu.cu.cc \ -I$TF_INC -I$HEADER -L$TF_LIB -D GOOGLE_CUDA=1 -x cu -Xcompiler -fPIC $CXXFLAGS --expt-relaxed-constexpr \ -arch=sm_37

g++ -std=c++11 -shared -ltensorflow_framework -o roi_pooling.so roi_pooling_op.cc \ roi_pooling_op.cu.o -I $TF_INC -L$TF_LIB -I $HEADER -D GOOGLE_CUDA=1 -fPIC $CXXFLAGS \ -lcudart -L $CUDA_PATH/lib else g++ -std=c++11 -I$HEADER -L$TF_LIB -ltensorflow_framework -shared -o roi_pooling.so roi_pooling_op.cc \ -I $TF_INC -fPIC $CXXFLAGS fi

cd ..