Something on RoIAlign --- basic introduction and implementation
2018-10-22 22:40:09
Paper: Mask RCNN
Code: https://github.com/longcw/RoIAlign.pytorch
Blog:
1. https://www.cnblogs.com/wangyong/p/8523814.html
2. https://blog.csdn.net/JNingWei/article/details/78822159
3. https://blog.csdn.net/Julialove102123/article/details/80567827
=========== Introduction ===========
see this blog: https://www.cnblogs.com/wangxiaocvpr/p/9840230.html
=========== Implementation ===========
git clone https://github.com/longcw/RoIAlign.pytorch
cd RoIAlign.pytorch
modify the script install.sh and test.sh into the following way:
#!/usr/bin/env bash
CUDA_PATH=/usr/local/cuda
echo "Compiling crop_and_resize kernels by nvcc..."
cd roi_align/src/cuda
$CUDA_PATH/bin/nvcc -c -o crop_and_resize_kernel.cu.o crop_and_resize_kernel.cu -x cu -Xcompiler -fPIC -arch=sm_30
-gencode=arch=compute_30,code=sm_30
-gencode=arch=compute_50,code=sm_50
-gencode=arch=compute_52,code=sm_52
-gencode=arch=compute_60,code=sm_60
-gencode=arch=compute_61,code=sm_61
-gencode=arch=compute_62,code=sm_62
cd ../../../roi_align
python3 build.py
cd ..
python3 setup.py install
#find $CONDA_PREFIX -name roi_align | awk '{mkdir $0 "/_ext" }' |bash
#find $CONDA_PREFIX -name roi_align | awk '{print "cp -r roi_align/_ext/* " $0 "/_ext/" }' |bash
python3 tests/test.py
python3 tests/test2.py
python3 tests/crop_and_resize_example.py
then, run the test.sh, you can found this:
wangxiao@AHU:/media/wangxiao/b8efbc67-7ea5-476d-9631-70da75f84e2d/reference_code/RoIAlign.pytorch$ sh ./test.sh
/usr/local/lib/python3.6/importlib/_bootstrap.py:219: RuntimeWarning: compiletime version 3.5 of module 'tensorflow.python.framework.fast_tensor_util' does not match runtime version 3.6
return f(*args, **kwds)
pytorch forward and backward start
pytorch forward and backward end
2018-10-22 22:38:28.483699: I tensorflow/core/platform/cpu_feature_guard.cc:137] Your CPU supports instructions that this TensorFlow binary was not compiled to use: SSE4.1 SSE4.2 AVX AVX2 FMA
2018-10-22 22:38:28.492705: I tensorflow/stream_executor/cuda/cuda_gpu_executor.cc:892] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero
2018-10-22 22:38:28.493019: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1030] Found device 0 with properties:
name: GeForce GTX 1080 major: 6 minor: 1 memoryClockRate(GHz): 1.898
pciBusID: 0000:02:00.0
totalMemory: 7.92GiB freeMemory: 6.29GiB
2018-10-22 22:38:28.493047: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1120] Creating TensorFlow device (/device:GPU:0) -> (device: 0, name: GeForce GTX 1080, pci bus id: 0000:02:00.0, compute capability: 6.1)
2018-10-22 22:38:28.737003: E tensorflow/stream_executor/cuda/cuda_dnn.cc:378] Loaded runtime CuDNN library: 7102 (compatibility version 7100) but source was compiled with 6021 (compatibility version 6000). If using a binary install, upgrade your CuDNN library to match. If building from sources, make sure the library loaded at runtime matches a compatible version specified during compile configuration.
2018-10-22 22:38:28.737107: F tensorflow/core/kernels/conv_ops.cc:667] Check failed: stream->parent()->GetConvolveAlgorithms( conv_parameters.ShouldIncludeWinogradNonfusedAlgo<T>(), &algorithms)
Aborted
tensor([[[[0., 1., 2.],
[0., 1., 2.],
[0., 1., 2.]]]], grad_fn=<CropAndResizeFunction>)
torch.Size([2, 3, 500, 500])
Some Bug you may meet:
1. cffi.error.VerificationError: LinkError: command 'x86_64-linux-gnu-gcc' failed with exit status 1
==>> run the followings in the terminal before you run "sh make.sh":
export CUDA_PATH=/usr/local/cuda/
export CXXFLAGS="-std=c++11"
export CFLAGS="-std=c99"
export PATH=/usr/local/cuda-8.0/bin${PATH:+:${PATH}}
export CPATH=/usr/local/cuda-8.0/include${CPATH:+:${CPATH}}
export LD_LIBRARY_PATH=/usr/local/cuda-8.0/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
then, it will shown you the followings:
wangxiao@AHU:~/Documents/Detectron.pytorch/lib$ sh make.sh
running build_ext
building 'utils.cython_bbox' extension
creating build
creating build/temp.linux-x86_64-2.7
creating build/temp.linux-x86_64-2.7/utils
x86_64-linux-gnu-gcc -pthread -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -std=c99 -fPIC -I/usr/local/lib/python2.7/dist-packages/numpy/core/include -I/usr/include/python2.7 -c utils/cython_bbox.c -o build/temp.linux-x86_64-2.7/utils/cython_bbox.o -Wno-cpp
creating build/lib.linux-x86_64-2.7
creating build/lib.linux-x86_64-2.7/utils
x86_64-linux-gnu-gcc -pthread -shared -Wl,-O1 -Wl,-Bsymbolic-functions -Wl,-Bsymbolic-functions -Wl,-z,relro -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -D_FORTIFY_SOURCE=2 -g -fstack-protector --param=ssp-buffer-size=4 -Wformat -Werror=format-security -std=c99 build/temp.linux-x86_64-2.7/utils/cython_bbox.o -o build/lib.linux-x86_64-2.7/utils/cython_bbox.so
building 'utils.cython_nms' extension
x86_64-linux-gnu-gcc -pthread -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -std=c99 -fPIC -I/usr/local/lib/python2.7/dist-packages/numpy/core/include -I/usr/include/python2.7 -c utils/cython_nms.c -o build/temp.linux-x86_64-2.7/utils/cython_nms.o -Wno-cpp
x86_64-linux-gnu-gcc -pthread -shared -Wl,-O1 -Wl,-Bsymbolic-functions -Wl,-Bsymbolic-functions -Wl,-z,relro -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -D_FORTIFY_SOURCE=2 -g -fstack-protector --param=ssp-buffer-size=4 -Wformat -Werror=format-security -std=c99 build/temp.linux-x86_64-2.7/utils/cython_nms.o -o build/lib.linux-x86_64-2.7/utils/cython_nms.so
copying build/lib.linux-x86_64-2.7/utils/cython_bbox.so -> utils
copying build/lib.linux-x86_64-2.7/utils/cython_nms.so -> utils
Compiling nms kernels by nvcc...
Including CUDA code.
/home/wangxiao/Documents/Detectron.pytorch/lib/model/nms
['/home/wangxiao/Documents/Detectron.pytorch/lib/model/nms/src/nms_cuda_kernel.cu.o']
generating /tmp/tmp1cFjY6/_nms.c
setting the current directory to '/tmp/tmp1cFjY6'
running build_ext
building '_nms' extension
creating home
creating home/wangxiao
creating home/wangxiao/Documents
creating home/wangxiao/Documents/Detectron.pytorch
creating home/wangxiao/Documents/Detectron.pytorch/lib
creating home/wangxiao/Documents/Detectron.pytorch/lib/model
creating home/wangxiao/Documents/Detectron.pytorch/lib/model/nms
creating home/wangxiao/Documents/Detectron.pytorch/lib/model/nms/src
x86_64-linux-gnu-gcc -pthread -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -std=c99 -fPIC -DWITH_CUDA -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/TH -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/THC -I/usr/local/cuda/include -I/usr/include/python2.7 -c _nms.c -o ./_nms.o
x86_64-linux-gnu-gcc -pthread -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -std=c99 -fPIC -DWITH_CUDA -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/TH -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/THC -I/usr/local/cuda/include -I/usr/include/python2.7 -c /home/wangxiao/Documents/Detectron.pytorch/lib/model/nms/src/nms_cuda.c -o ./home/wangxiao/Documents/Detectron.pytorch/lib/model/nms/src/nms_cuda.o
x86_64-linux-gnu-gcc -pthread -shared -Wl,-O1 -Wl,-Bsymbolic-functions -Wl,-Bsymbolic-functions -Wl,-z,relro -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -D_FORTIFY_SOURCE=2 -g -fstack-protector --param=ssp-buffer-size=4 -Wformat -Werror=format-security -std=c99 ./_nms.o ./home/wangxiao/Documents/Detectron.pytorch/lib/model/nms/src/nms_cuda.o /home/wangxiao/Documents/Detectron.pytorch/lib/model/nms/src/nms_cuda_kernel.cu.o -o ./_nms.so
Compiling roi pooling kernels by nvcc...
Including CUDA code.
/home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_pooling
generating /tmp/tmpUUPhTw/_roi_pooling.c
setting the current directory to '/tmp/tmpUUPhTw'
running build_ext
building '_roi_pooling' extension
creating home
creating home/wangxiao
creating home/wangxiao/Documents
creating home/wangxiao/Documents/Detectron.pytorch
creating home/wangxiao/Documents/Detectron.pytorch/lib
creating home/wangxiao/Documents/Detectron.pytorch/lib/model
creating home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_pooling
creating home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_pooling/src
x86_64-linux-gnu-gcc -pthread -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -std=c99 -fPIC -DWITH_CUDA -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/TH -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/THC -I/usr/local/cuda/include -I/usr/include/python2.7 -c _roi_pooling.c -o ./_roi_pooling.o
x86_64-linux-gnu-gcc -pthread -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -std=c99 -fPIC -DWITH_CUDA -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/TH -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/THC -I/usr/local/cuda/include -I/usr/include/python2.7 -c /home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_pooling/src/roi_pooling.c -o ./home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_pooling/src/roi_pooling.o
x86_64-linux-gnu-gcc -pthread -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -std=c99 -fPIC -DWITH_CUDA -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/TH -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/THC -I/usr/local/cuda/include -I/usr/include/python2.7 -c /home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_pooling/src/roi_pooling_cuda.c -o ./home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_pooling/src/roi_pooling_cuda.o
x86_64-linux-gnu-gcc -pthread -shared -Wl,-O1 -Wl,-Bsymbolic-functions -Wl,-Bsymbolic-functions -Wl,-z,relro -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -D_FORTIFY_SOURCE=2 -g -fstack-protector --param=ssp-buffer-size=4 -Wformat -Werror=format-security -std=c99 ./_roi_pooling.o ./home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_pooling/src/roi_pooling.o ./home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_pooling/src/roi_pooling_cuda.o /home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_pooling/src/roi_pooling.cu.o -o ./_roi_pooling.so
Compiling roi crop kernels by nvcc...
Including CUDA code.
/home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_crop
generating /tmp/tmp2NBTVb/_roi_crop.c
setting the current directory to '/tmp/tmp2NBTVb'
running build_ext
building '_roi_crop' extension
creating home
creating home/wangxiao
creating home/wangxiao/Documents
creating home/wangxiao/Documents/Detectron.pytorch
creating home/wangxiao/Documents/Detectron.pytorch/lib
creating home/wangxiao/Documents/Detectron.pytorch/lib/model
creating home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_crop
creating home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_crop/src
x86_64-linux-gnu-gcc -pthread -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -std=c99 -fPIC -DWITH_CUDA -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/TH -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/THC -I/usr/local/cuda/include -I/usr/include/python2.7 -c _roi_crop.c -o ./_roi_crop.o
x86_64-linux-gnu-gcc -pthread -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -std=c99 -fPIC -DWITH_CUDA -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/TH -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/THC -I/usr/local/cuda/include -I/usr/include/python2.7 -c /home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_crop/src/roi_crop.c -o ./home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_crop/src/roi_crop.o
/home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_crop/src/roi_crop.c: In function ‘BilinearSamplerBHWD_updateGradInput’:
/home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_crop/src/roi_crop.c:190:14: warning: unused variable ‘inBottomRight’ [-Wunused-variable]
real inBottomRight=0;
^
/home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_crop/src/roi_crop.c:189:14: warning: unused variable ‘inBottomLeft’ [-Wunused-variable]
real inBottomLeft=0;
^
/home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_crop/src/roi_crop.c:188:14: warning: unused variable ‘inTopRight’ [-Wunused-variable]
real inTopRight=0;
^
/home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_crop/src/roi_crop.c:187:14: warning: unused variable ‘inTopLeft’ [-Wunused-variable]
real inTopLeft=0;
^
/home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_crop/src/roi_crop.c:186:14: warning: unused variable ‘v’ [-Wunused-variable]
real v=0;
^
/home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_crop/src/roi_crop.c: In function ‘BilinearSamplerBCHW_updateGradInput’:
/home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_crop/src/roi_crop.c:440:14: warning: unused variable ‘inBottomRight’ [-Wunused-variable]
real inBottomRight=0;
^
/home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_crop/src/roi_crop.c:439:14: warning: unused variable ‘inBottomLeft’ [-Wunused-variable]
real inBottomLeft=0;
^
/home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_crop/src/roi_crop.c:438:14: warning: unused variable ‘inTopRight’ [-Wunused-variable]
real inTopRight=0;
^
/home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_crop/src/roi_crop.c:437:14: warning: unused variable ‘inTopLeft’ [-Wunused-variable]
real inTopLeft=0;
^
/home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_crop/src/roi_crop.c:436:14: warning: unused variable ‘v’ [-Wunused-variable]
real v=0;
^
x86_64-linux-gnu-gcc -pthread -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -std=c99 -fPIC -DWITH_CUDA -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/TH -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/THC -I/usr/local/cuda/include -I/usr/include/python2.7 -c /home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_crop/src/roi_crop_cuda.c -o ./home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_crop/src/roi_crop_cuda.o
x86_64-linux-gnu-gcc -pthread -shared -Wl,-O1 -Wl,-Bsymbolic-functions -Wl,-Bsymbolic-functions -Wl,-z,relro -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -D_FORTIFY_SOURCE=2 -g -fstack-protector --param=ssp-buffer-size=4 -Wformat -Werror=format-security -std=c99 ./_roi_crop.o ./home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_crop/src/roi_crop.o ./home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_crop/src/roi_crop_cuda.o /home/wangxiao/Documents/Detectron.pytorch/lib/model/roi_crop/src/roi_crop_cuda_kernel.cu.o -o ./_roi_crop.so
Compiling roi align kernels by nvcc...
Including CUDA code.
/home/wangxiao/Documents/Detectron.pytorch/lib/modeling/roi_xfrom/roi_align
generating /tmp/tmptraHMG/_roi_align.c
setting the current directory to '/tmp/tmptraHMG'
running build_ext
building '_roi_align' extension
creating home
creating home/wangxiao
creating home/wangxiao/Documents
creating home/wangxiao/Documents/Detectron.pytorch
creating home/wangxiao/Documents/Detectron.pytorch/lib
creating home/wangxiao/Documents/Detectron.pytorch/lib/modeling
creating home/wangxiao/Documents/Detectron.pytorch/lib/modeling/roi_xfrom
creating home/wangxiao/Documents/Detectron.pytorch/lib/modeling/roi_xfrom/roi_align
creating home/wangxiao/Documents/Detectron.pytorch/lib/modeling/roi_xfrom/roi_align/src
x86_64-linux-gnu-gcc -pthread -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -std=c99 -fPIC -DWITH_CUDA -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/TH -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/THC -I/usr/local/cuda/include -I/usr/include/python2.7 -c _roi_align.c -o ./_roi_align.o
x86_64-linux-gnu-gcc -pthread -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -std=c99 -fPIC -DWITH_CUDA -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/TH -I/usr/local/lib/python2.7/dist-packages/torch/utils/ffi/../../lib/include/THC -I/usr/local/cuda/include -I/usr/include/python2.7 -c /home/wangxiao/Documents/Detectron.pytorch/lib/modeling/roi_xfrom/roi_align/src/roi_align_cuda.c -o ./home/wangxiao/Documents/Detectron.pytorch/lib/modeling/roi_xfrom/roi_align/src/roi_align_cuda.o
x86_64-linux-gnu-gcc -pthread -shared -Wl,-O1 -Wl,-Bsymbolic-functions -Wl,-Bsymbolic-functions -Wl,-z,relro -fno-strict-aliasing -DNDEBUG -g -fwrapv -O2 -Wall -Wstrict-prototypes -D_FORTIFY_SOURCE=2 -g -fstack-protector --param=ssp-buffer-size=4 -Wformat -Werror=format-security -std=c99 ./_roi_align.o ./home/wangxiao/Documents/Detectron.pytorch/lib/modeling/roi_xfrom/roi_align/src/roi_align_cuda.o /home/wangxiao/Documents/Detectron.pytorch/lib/modeling/roi_xfrom/roi_align/src/roi_align_kernel.cu.o -o ./_roi_align.so
wangxiao@AHU:~/Documents/Detectron.pytorch/lib$
2. You may also meet the following issues: AttributeError: 'module' object has no attribute 'getLogger'
Traceback (most recent call last):
File "tools/train_net_step.py", line 22, in <module>
import utils.net as net_utils
File "/home/Detectron.pytorch/lib/utils/net.py", line 12, in <module>
logger = logging.getLogger(__name__)
AttributeError: 'module' object has no attribute 'getLogger'
you need to re-compile the lib with python3.6. i.e. change the make.sh into follows:
#!/usr/bin/env bash
CUDA_PATH=/usr/local/cuda/
python3 setup.py build_ext --inplace
rm -rf build
# Choose cuda arch as you need
CUDA_ARCH="-gencode arch=compute_30,code=sm_30
-gencode arch=compute_35,code=sm_35
-gencode arch=compute_50,code=sm_50
-gencode arch=compute_52,code=sm_52
-gencode arch=compute_60,code=sm_60
-gencode arch=compute_61,code=sm_61 "
# -gencode arch=compute_70,code=sm_70 "
# compile NMS
cd model/nms/src
echo "Compiling nms kernels by nvcc..."
nvcc -c -o nms_cuda_kernel.cu.o nms_cuda_kernel.cu
-D GOOGLE_CUDA=1 -x cu -Xcompiler -fPIC $CUDA_ARCH
cd ../
python3 build.py
# compile roi_pooling
cd ../../
cd model/roi_pooling/src
echo "Compiling roi pooling kernels by nvcc..."
nvcc -c -o roi_pooling.cu.o roi_pooling_kernel.cu
-D GOOGLE_CUDA=1 -x cu -Xcompiler -fPIC $CUDA_ARCH
cd ../
python3 build.py
# # compile roi_align
# cd ../../
# cd model/roi_align/src
# echo "Compiling roi align kernels by nvcc..."
# nvcc -c -o roi_align_kernel.cu.o roi_align_kernel.cu
# -D GOOGLE_CUDA=1 -x cu -Xcompiler -fPIC $CUDA_ARCH
# cd ../
# python3 build.py
# compile roi_crop
cd ../../
cd model/roi_crop/src
echo "Compiling roi crop kernels by nvcc..."
nvcc -c -o roi_crop_cuda_kernel.cu.o roi_crop_cuda_kernel.cu
-D GOOGLE_CUDA=1 -x cu -Xcompiler -fPIC $CUDA_ARCH
cd ../
python3 build.py
# compile roi_align (based on Caffe2's implementation)
cd ../../
cd modeling/roi_xfrom/roi_align/src
echo "Compiling roi align kernels by nvcc..."
nvcc -c -o roi_align_kernel.cu.o roi_align_kernel.cu
-D GOOGLE_CUDA=1 -x cu -Xcompiler -fPIC $CUDA_ARCH
cd ../
python3 build.py
you also need to copy the file: e2e_mask_rcnn_R-50-C4_1x.yaml and rename it as: e2e_mask_rcnn_R-50-C4.yml
then, you can run the script to train the model:
$ python3 tools/train_net_step.py --dataset coco2017 --cfg configs/baselines/e2e_mask_rcnn_R-50-C4.yml --use_tfboard --bs 5 --nw 2
wangxiao@AHU:~/Documents/Detectron.pytorch$ python3 tools/train_net_step.py --dataset coco2017 --cfg configs/baselines/e2e_mask_rcnn_R-50-C4.yml --use_tfboard --bs 1 --nw 1
Called with args:
Namespace(batch_size=1, cfg_file='configs/baselines/e2e_mask_rcnn_R-50-C4.yml', cuda=True, dataset='coco2017', disp_interval=20, iter_size=1, load_ckpt=None, load_detectron=None, lr=None, lr_decay_gamma=None, no_save=False, num_workers=1, optimizer=None, resume=False, set_cfgs=[], start_step=0, use_tfboard=True)
effective_batch_size = batch_size * iter_size = 1 * 1
Adaptive config changes:
effective_batch_size: 8 --> 1
NUM_GPUS: 8 --> 1
IMS_PER_BATCH: 1 --> 1
Adjust BASE_LR linearly according to batch_size change:
BASE_LR: 0.01 --> 0.00125
Adjust SOLVER.STEPS and SOLVER.MAX_ITER linearly based on effective_batch_size change:
SOLVER.STEPS: [0, 120000, 160000] --> [0, 960000, 1280000]
SOLVER.MAX_ITER: 180000 --> 1440000
Number of data loading threads: 1
loading annotations into memory...
Done (t=9.39s)
creating index...
index created!
INFO json_dataset.py: 298: Loading cached gt_roidb from /home/wangxiao/Documents/Detectron.pytorch/data/cache/coco_2017_train_gt_roidb.pkl
INFO roidb.py: 50: Appending horizontally-flipped training examples...
INFO roidb.py: 52: Loaded dataset: coco_2017_train
INFO roidb.py: 143: Filtered 2042 roidb entries: 236574 -> 234532
INFO roidb.py: 69: Computing image aspect ratios and ordering the ratios...
INFO roidb.py: 71: done
INFO roidb.py: 75: Computing bounding-box regression targets...
INFO roidb.py: 77: done
INFO train_net_step.py: 233: 234532 roidb entries
INFO train_net_step.py: 234: Takes 93.39 sec(s) to construct roidb
INFO train_net_step.py: 378: Training starts !
INFO net.py: 72: Changing learning rate 0.000000 -> 0.000417
/usr/local/lib/python3.6/site-packages/torch/nn/functional.py:1006: UserWarning: nn.functional.sigmoid is deprecated. Use torch.sigmoid instead.
warnings.warn("nn.functional.sigmoid is deprecated. Use torch.sigmoid instead.")
/usr/local/lib/python3.6/site-packages/torch/nn/functional.py:52: UserWarning: size_average and reduce args will be deprecated, please use reduction='sum' instead.
warnings.warn(warning.format(ret))
[Nov24-16-46-26_AHU_step][e2e_mask_rcnn_R-50-C4.yml][Step 1 / 1440000]
loss: 6.916170, lr: 0.000417 time: 0.721983, eta: 12 days, 0:47:35
accuracy_cls: 0.001953
loss_cls: 4.648660, loss_bbox: 0.030453, loss_mask: 0.858463
loss_rpn_cls: 0.676336, loss_rpn_bbox: 0.702258
==