Revert "Porting this code to Windows"

.gitignore

@@ -1,17 +1,3 @@
 data/*
 *.pyc
 *~
-/images - Copy
-*.pyd
-*.ipch
-*.tlog
-*.deps
-*.cache
-*.obj
-*.log
-*.pdb
-/vgg16_caffe.pth
-/windows/.vs/faster_rcnn/v15
-*.dll
-*.lib
-/.vs/FasterRCNN/v15

README.md

@@ -149,10 +149,8 @@ Install all the python dependencies using pip:
 pip install -r requirements.txt
 ```
 
-if bulid on windows, please click [README.md](./windows/README.md).
-
-else compile the cuda dependencies using following simple commands:
-
+Compile the cuda dependencies using following simple commands:
+
 ```
 cd lib
 sh make.sh

demo.py

@@ -55,13 +55,13 @@ def parse_args():
                       default='cfgs/vgg16.yml', type=str)
   parser.add_argument('--net', dest='net',
                       help='vgg16, res50, res101, res152',
-                      default='vgg16', type=str)
+                      default='res101', type=str)
   parser.add_argument('--set', dest='set_cfgs',
                       help='set config keys', default=None,
                       nargs=argparse.REMAINDER)
   parser.add_argument('--load_dir', dest='load_dir',
                       help='directory to load models',
-                      default="F:/Pascal/faster_rcnn_pytorch")
+                      default="/srv/share/jyang375/models")
   parser.add_argument('--image_dir', dest='image_dir',
                       help='directory to load images for demo',
                       default="images")

lib/make.sh

@@ -1,6 +1,6 @@
 #!/usr/bin/env bash
 
-CUDA_PATH="A:/NVIDIA/GPU Computing Toolkit/CUDA/v9.1"
+CUDA_PATH=/usr/local/cuda/
 
 python setup.py build_ext --inplace
 rm -rf build
@@ -15,8 +15,8 @@ CUDA_ARCH="-gencode arch=compute_30,code=sm_30 \
 # compile NMS
 cd model/nms/src
 echo "Compiling nms kernels by nvcc..."
-#nvcc -c -o nms_cuda_kernel.lib nms_cuda_kernel.cu \
-#     -D GOOGLE_CUDA=1 -x cu -Xcompiler -fPIC $CUDA_ARCH
+nvcc -c -o nms_cuda_kernel.cu.o nms_cuda_kernel.cu \
+	 -D GOOGLE_CUDA=1 -x cu -Xcompiler -fPIC $CUDA_ARCH
 
 cd ../
 python build.py
@@ -25,25 +25,25 @@ python build.py
 cd ../../
 cd model/roi_pooling/src
 echo "Compiling roi pooling kernels by nvcc..."
-#nvcc -c -o roi_pooling.lib roi_pooling_kernel.cu \
-#     -D GOOGLE_CUDA=1 -x cu -Xcompiler -fPIC $CUDA_ARCH
+nvcc -c -o roi_pooling.cu.o roi_pooling_kernel.cu \
+	 -D GOOGLE_CUDA=1 -x cu -Xcompiler -fPIC $CUDA_ARCH
 cd ../
 python build.py
 
 # compile roi_align
 cd ../../
 cd model/roi_align/src
 echo "Compiling roi align kernels by nvcc..."
-#nvcc -c -o roi_align_kernel.lib roi_align_kernel.cu \
-#	  -D GOOGLE_CUDA=1 -x cu -Xcompiler -fPIC $CUDA_ARCH
+nvcc -c -o roi_align_kernel.cu.o roi_align_kernel.cu \
+	 -D GOOGLE_CUDA=1 -x cu -Xcompiler -fPIC $CUDA_ARCH
 cd ../
 python 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.lib roi_crop_cuda_kernel.cu \
-#     -D GOOGLE_CUDA=1 -x cu -Xcompiler -fPIC $CUDA_ARCH
+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 ../
 python build.py

lib/model/nms/_ext/nms/__init__.py

@@ -6,10 +6,7 @@
 def _import_symbols(locals):
     for symbol in dir(_lib):
         fn = getattr(_lib, symbol)
-        if callable(fn):
-            locals[symbol] = _wrap_function(fn, _ffi)
-        else:
-            locals[symbol] = fn
+        locals[symbol] = _wrap_function(fn, _ffi)
         __all__.append(symbol)
 
 _import_symbols(locals())

lib/model/nms/build.py

@@ -1,36 +1,25 @@
 from __future__ import print_function
 import os
-import sys
 import torch
 from torch.utils.ffi import create_extension
 
 #this_file = os.path.dirname(__file__)
-torch_root = os.path.join(os.path.dirname(sys.executable),
-                          'Lib/site-packages/torch/lib')
-cuda_root = os.environ['CUDA_PATH']
 
 sources = []
 headers = []
 defines = []
-include_dirs = []
 with_cuda = False
 
-this_file = os.path.dirname(os.path.realpath(__file__))
-print(this_file)
-
 if torch.cuda.is_available():
     print('Including CUDA code.')
-    sources += ['src/nms_cuda.cpp']
+    sources += ['src/nms_cuda.c']
     headers += ['src/nms_cuda.h']
-    include_dirs += [os.path.join(cuda_root,"include"), os.path.join(torch_root,'include')]
     defines += [('WITH_CUDA', None)]
     with_cuda = True
 
-
-extra_objects =  ['src/nms_cuda_kernel.lib']
-extra_objects += [os.path.join(torch_root,'ATen.lib'),
-                  os.path.join(cuda_root,'lib/x64/cudart.lib'),
-                  os.path.join(torch_root,'_C.lib')]
+this_file = os.path.dirname(os.path.realpath(__file__))
+print(this_file)
+extra_objects = ['src/nms_cuda_kernel.cu.o']
 extra_objects = [os.path.join(this_file, fname) for fname in extra_objects]
 print(extra_objects)
 
@@ -41,8 +30,7 @@
     define_macros=defines,
     relative_to=__file__,
     with_cuda=with_cuda,
-    extra_objects=extra_objects,
-    include_dirs=include_dirs
+    extra_objects=extra_objects
 )
 
 if __name__ == '__main__':

lib/model/nms/src/nms_cuda.c

@@ -0,0 +1,19 @@
+#include <THC/THC.h>
+#include <stdio.h>
+#include "nms_cuda_kernel.h"
+
+// this symbol will be resolved automatically from PyTorch libs
+extern THCState *state;
+
+int nms_cuda(THCudaIntTensor *keep_out, THCudaTensor *boxes_host,
+		     THCudaIntTensor *num_out, float nms_overlap_thresh) {
+
+	nms_cuda_compute(THCudaIntTensor_data(state, keep_out), 
+		         THCudaIntTensor_data(state, num_out), 
+      	                 THCudaTensor_data(state, boxes_host), 
+		         boxes_host->size[0], 
+		         boxes_host->size[1],
+		         nms_overlap_thresh);
+
+	return 1;
+}

lib/model/nms/src/nms_cuda.h

@@ -1,4 +1,5 @@
 // int nms_cuda(THCudaTensor *keep_out, THCudaTensor *num_out,
 //             THCudaTensor *boxes_host, THCudaTensor *nms_overlap_thresh);
+
 int nms_cuda(THCudaIntTensor *keep_out, THCudaTensor *boxes_host,
              THCudaIntTensor *num_out, float nms_overlap_thresh);

lib/model/roi_align/_ext/roi_align/__init__.py

@@ -6,10 +6,7 @@
 def _import_symbols(locals):
     for symbol in dir(_lib):
         fn = getattr(_lib, symbol)
-        if callable(fn):
-            locals[symbol] = _wrap_function(fn, _ffi)
-        else:
-            locals[symbol] = fn
+        locals[symbol] = _wrap_function(fn, _ffi)
         __all__.append(symbol)
 
 _import_symbols(locals())

lib/model/roi_align/build.py

@@ -1,17 +1,13 @@
 from __future__ import print_function
 import os
-import sys
 import torch
 from torch.utils.ffi import create_extension
 
-
-torch_root = os.path.join(os.path.dirname(sys.executable),
-                          'Lib/site-packages/torch/lib')
-cuda_root = os.environ['CUDA_PATH']
-
-sources = ['src/roi_align.cpp']
+sources = ['src/roi_align.c']
 headers = ['src/roi_align.h']
-include_dirs = []
+extra_objects = []
+#sources = []
+#headers = []
 defines = []
 with_cuda = False
 
@@ -20,18 +16,13 @@
 
 if torch.cuda.is_available():
     print('Including CUDA code.')
-    sources += ['src/roi_align_cuda.cpp']
+    sources += ['src/roi_align_cuda.c']
     headers += ['src/roi_align_cuda.h']
     defines += [('WITH_CUDA', None)]
     with_cuda = True
-    include_dirs += [os.path.join(cuda_root,"include"), 
-                     os.path.join(torch_root,'include')]
-extra_objects = ['src/roi_align_kernel.lib']
-extra_objects = [os.path.join(this_file, fname) for fname in extra_objects]
-extra_objects += [os.path.join(torch_root,'ATen.lib'),
-                  os.path.join(cuda_root,'lib/x64/cudart.lib'),
-                  os.path.join(torch_root,'_C.lib')]
-print(extra_objects)
+
+    extra_objects = ['src/roi_align_kernel.cu.o']
+    extra_objects = [os.path.join(this_file, fname) for fname in extra_objects]
 
 ffi = create_extension(
     '_ext.roi_align',
@@ -40,8 +31,7 @@
     define_macros=defines,
     relative_to=__file__,
     with_cuda=with_cuda,
-    extra_objects=extra_objects,
-    include_dirs=include_dirs
+    extra_objects=extra_objects
 )
 
 if __name__ == '__main__':

lib/model/roi_align/src/roi_align.cpp → lib/model/roi_align/src/roi_align.c

@@ -13,7 +13,7 @@ void ROIAlignBackwardCpu(const float* top_diff, const float spatial_scale, const
                      const int aligned_height, const int aligned_width, const float * bottom_rois,
                      float* top_data);
 
-extern"C" __declspec(dllexport) int roi_align_forward(int aligned_height, int aligned_width, float spatial_scale,
+int roi_align_forward(int aligned_height, int aligned_width, float spatial_scale,
                      THFloatTensor * features, THFloatTensor * rois, THFloatTensor * output)
 {
     //Grab the input tensor
@@ -44,7 +44,7 @@ extern"C" __declspec(dllexport) int roi_align_forward(int aligned_height, int al
     return 1;
 }
 
-extern"C" __declspec(dllexport) int roi_align_backward(int aligned_height, int aligned_width, float spatial_scale,
+int roi_align_backward(int aligned_height, int aligned_width, float spatial_scale,
                        THFloatTensor * top_grad, THFloatTensor * rois, THFloatTensor * bottom_grad)
 {
     //Grab the input tensor

lib/model/roi_align/src/roi_align_cuda.cpp → lib/model/roi_align/src/roi_align_cuda.c

@@ -1,11 +1,10 @@
 #include <THC/THC.h>
 #include <math.h>
-#include <ATen/ATen.h>
 #include "roi_align_kernel.h"
 
-THCState *state = at::globalContext().thc_state;
+extern THCState *state;
 
-extern"C" __declspec(dllexport) int roi_align_forward_cuda(int aligned_height, int aligned_width, float spatial_scale,
+int roi_align_forward_cuda(int aligned_height, int aligned_width, float spatial_scale,
                         THCudaTensor * features, THCudaTensor * rois, THCudaTensor * output)
 {
     // Grab the input tensor
@@ -40,7 +39,7 @@ extern"C" __declspec(dllexport) int roi_align_forward_cuda(int aligned_height, i
     return 1;
 }
 
-extern"C" __declspec(dllexport) int roi_align_backward_cuda(int aligned_height, int aligned_width, float spatial_scale,
+int roi_align_backward_cuda(int aligned_height, int aligned_width, float spatial_scale,
                         THCudaTensor * top_grad, THCudaTensor * rois, THCudaTensor * bottom_grad)
 {
     // Grab the input tensor

lib/model/roi_crop/_ext/roi_crop/__init__.py

@@ -6,10 +6,7 @@
 def _import_symbols(locals):
     for symbol in dir(_lib):
         fn = getattr(_lib, symbol)
-        if callable(fn):
-            locals[symbol] = _wrap_function(fn, _ffi)
-        else:
-            locals[symbol] = fn
+        locals[symbol] = _wrap_function(fn, _ffi)
         __all__.append(symbol)
 
 _import_symbols(locals())

lib/model/roi_crop/build.py

@@ -1,38 +1,26 @@
 from __future__ import print_function
 import os
-import sys
 import torch
 from torch.utils.ffi import create_extension
 
 #this_file = os.path.dirname(__file__)
-torch_root = os.path.join(os.path.dirname(sys.executable),
-                          'Lib/site-packages/torch/lib')
-cuda_root = os.environ['CUDA_PATH']
 
-sources = ['src/roi_crop.cpp']
+sources = ['src/roi_crop.c']
 headers = ['src/roi_crop.h']
-include_dirs = []
 defines = []
 with_cuda = False
 
-this_file = os.path.dirname(os.path.realpath(__file__))
-print(this_file)
-
 if torch.cuda.is_available():
     print('Including CUDA code.')
-    sources += ['src/roi_crop_cuda.cpp']
+    sources += ['src/roi_crop_cuda.c']
     headers += ['src/roi_crop_cuda.h']
-    include_dirs += [os.path.join(cuda_root,"include"), 
-                     os.path.join(torch_root,'include')]
     defines += [('WITH_CUDA', None)]
     with_cuda = True
 
-extra_objects = ['src/roi_crop_cuda_kernel.lib']
+this_file = os.path.dirname(os.path.realpath(__file__))
+print(this_file)
+extra_objects = ['src/roi_crop_cuda_kernel.cu.o']
 extra_objects = [os.path.join(this_file, fname) for fname in extra_objects]
-extra_objects += [os.path.join(torch_root,'ATen.lib'),
-                  os.path.join(cuda_root,'lib/x64/cudart.lib'),
-                  os.path.join(torch_root,'_C.lib')]
-print(extra_objects)
 
 ffi = create_extension(
     '_ext.roi_crop',
@@ -41,8 +29,7 @@
     define_macros=defines,
     relative_to=__file__,
     with_cuda=with_cuda,
-    extra_objects=extra_objects,
-    include_dirs=include_dirs
+    extra_objects=extra_objects
 )
 
 if __name__ == '__main__':