faster-rcnn: multithreading for caffe:pooling_layer,relu_layer,roi_pooling_layer,im2col

8 threads CPU(i7-6700)layer speed up x5 ~ x7,(im2col(x2.5) DDR4 is the bottlenecks)

• org code from  https://github.com/rbgirshick/py-faster-rcnn.git

Add fucntion

• add multithreading for caffe:

  modified: src/caffe/layers/pooling_layer.cpp

  modified: src/caffe/layers/relu_layer.cpp

  modified: src/caffe/layers/roi_pooling_layer.cpp

  modified: src/caffe/net.cpp

  modified: src/caffe/util/im2col.cpp

  add: include/caffe/calcu_pthread.h

  add: src/caffe/calcu_pthread.cpp

  add: Makefile.config

• add westwell port project using modle:

  modified: tools/demo.py

  add: westwell/

Compile:

• cd multithreading_frcnn/caffe-fast-rcnn

  make -j8 && make pycaffe

• cd multithreading_frcnn/lib

  make

Exec:

python ./tools/demo.py --cpu